Associations of Skeletal Muscle Mass and Percent Body Fat with Balance, Gait, and Locomotive Syndrome Risk in Community-dwelling Older Women
2026 年 11 巻 論文ID: 20260010
詳細
2026 年 11 巻 論文ID: 20260010
Objectives: Locomotive syndrome (LS) is prevalent among older adults, particularly women, and is a major contributor to the need for long-term care. In rapidly aging Japan, addressing LS has become a public health priority. Although skeletal muscle mass, percent body fat (PBF), balance, and walking speed have each been linked to LS, their interrelationships remain unclear. This study examined how muscle mass and PBF are associated with LS risk in relation to balance and walking speed.
Methods: This cross-sectional study analyzed community-dwelling older women aged 65 years or older, recruited from a larger cohort of adults aged 18 years or older who were able to ambulate independently. Assessments included skeletal muscle index (SMI), PBF, balance (single-leg standing time), walking speed, and LS status (two-step test, stand-up test, and the 25-question Geriatric Locomotive Function Scale). Path analysis was performed to examine the associations among variables.
Results: A total of 285 participants were included. The path model demonstrated acceptable fit (goodness-of-fit index = 0.986; adjusted goodness-of-fit index = 0.945; root mean square error of approximation = 0.056). PBF was associated with balance function (β = −0.24), which was statistically related to both walking speed (β = 0.33) and LS risk (β = −0.37). SMI showed no association with balance.
Conclusions: In predominantly non-sarcopenic older women, higher PBF was associated with increased LS risk, with these associations statistically related to balance and gait functions. Prospective studies are warranted to confirm these pathways and clarify causal relationships.
Locomotive syndrome (LS) is a major cause of long-term care needs and is particularly prevalent among older women.1) Women generally have lower skeletal muscle mass and higher body fat content than men, factors that may contribute to increased LS risk.2,3) Previous studies have also shown that muscle and fat mass affect physical function differently in men and women, highlighting the need to focus on women who are at greater risk of developing LS.4,5) In Japan, the population is aging rapidly because of increased average life expectancy and declining birth rates.6) As of 2022, the average life expectancy for Japanese women was 87.09 years, and more than a quarter (29.1%) of the Japanese population was aged 65 years or older.6) Consequently, the proportion of older adults requiring nursing care is expected to rise. To address this issue, the Japanese Orthopaedic Association (JOA) introduced the concept of LS to describe individuals with musculoskeletal conditions that increase the likelihood of requiring long-term care.7) LS is classified into three stages (1–3) according to severity, with higher stages indicating greater impairment.8) In Japan’s largest longitudinal study of older adults, the prevalence of LS among community-dwelling individuals was 69.8%, even at Stage 1, the mildest level.9) Given ongoing demographic shifts, the prevalence of LS is anticipated to continue increasing.
LS is associated with both sarcopenia (low skeletal muscle mass) and high percent body fat (PBF).10,11) Sarcopenia is a geriatric syndrome characterized by age-related declines in skeletal muscle mass, muscle strength, and physical performance.3) Clinical features of sarcopenia overlap with LS and influence LS progression.10,12) Among older Japanese women, a PBF of 37.3% or higher has been linked to an elevated risk of developing LS.11) An increase in PBF can alter body composition, which may impair balance function and increase LS risk.13,14) The open-eye single-leg standing (SLS) test, which evaluates balance function, is considered an effective screening tool for LS.15) Additionally, LS has been noted to be associated with decreased walking speed more so than with overall physical robustness.16) Therefore, a comprehensive assessment of skeletal muscle mass, PBF, balance function, and gait function may help identify individuals at risk for LS.
Despite these insights, no studies have examined the combined effects of skeletal muscle mass, PBF, balance function, and gait function on LS risk. Although each factor has been independently associated with LS, a comprehensive understanding of their interrelationships remains limited. The JOA currently recommends the stand-up test, two-step test, and the 25-question Geriatric Locomotive Function Scale (GLFS-25) to assess LS risk.17) If low skeletal muscle mass and high PBF are found to be associated with LS risk in relation to balance and gait function, this would suggest a novel pattern of associations relevant to LS. Furthermore, identifying such association patterns may help inform future approaches for preventing LS progression.
Structural equation modeling (SEM) is an effective analytical approach for exploring the complex interrelationships underlying LS risk. Because LS is influenced by multiple interrelated factors, analyses based on single variables may not adequately capture the complex pathways leading to its onset. SEM enables the simultaneous estimation of multiple interrelationships, providing a more comprehensive understanding of how skeletal muscle mass and PBF are associated with LS risk in relation to balance and gait functions.18)
This study aimed to examine the associations of skeletal muscle mass and PBF with LS risk in relation to balance and gait functions among community-dwelling older women. We hypothesized that lower skeletal muscle mass and higher PBF would be associated with lower balance and gait function, which are related to greater LS risk. Furthermore, this study seeks to provide insights that may inform future preventive approaches by clarifying association patterns involving both skeletal muscle mass and PBF.
This study used a cross-sectional design. Data on body composition, motor function, and LS were collected on the same day. A model was developed to examine hypothesized associations among these variables, and path analysis was performed to test whether the indirect associations were consistent with a mediation hypothesis. This study was approved by the Ethical Committee for Epidemiology of Hiroshima University (approval number: E2022-0086) and was conducted in accordance with the Declaration of Helsinki.
SettingData were obtained from the Study for Diagnosis, Early detection, and Treatment of locomotive syndrome using the Epidemiological Cohort (DETECt-L) database. Participants were recruited using explanatory posters between April and December 2023 in Hiroshima, Higashi-Hiroshima, and Kure, Japan. Data collection was conducted in gymnasiums and meeting rooms in sports and regional community centers.
ParticipantsThe following inclusion criteria were used: (1) community-dwelling adults aged 18 years or over and (2) those with independent mobility. Participants were excluded based on the following criteria: (1) suspected cognitive dysfunction and (2) serious medical conditions, including unstable cardiovascular disease, stroke, severe respiratory disease, Parkinson’s disease, diabetic peripheral neuropathy, or rheumatoid arthritis. For this study, only data from women aged 65 years or older were analyzed. A flyer outlining the study’s purpose and procedures was distributed to staff at the sports and regional community centers to assist with participant recruitment. Written informed consent was obtained from all participants.
VariablesThis study collected data on skeletal muscle mass, PBF, balance function, gait function, and LS risk. All assessments were conducted by licensed physical therapists and trained exercise instructors. Body composition was measured using a bioelectrical impedance analyzer (InBody 270; Inbody, Seoul, Korea) to assess skeletal muscle mass and PBF. The skeletal muscle index (SMI), a standard indicator of skeletal muscle mass in the assessment of sarcopenia, was calculated by dividing skeletal muscle mass (kg) by height squared (m2).
Grip strength was measured using a handgrip dynamometer (TKK 5401 Grip-D; Takei, Niigata, Japan). The muscle strength of each hand was measured in kilograms, and the average of the two values was used. Participants grasped the handgrip dynamometer with the pointer facing outward, and the interphalangeal joints of the fingers were positioned at approximately right angles before measurement. A previous study has demonstrated the high reliability and validity of the grip strength measurement method.19)
To assess balance function, SLS time was measured as the longest duration a participant could maintain balance on only one lower extremity. Before measurement, each participant selected the lower extremity that would hold the standing position. Participants were instructed to keep one lower extremity lifted off the floor during the single-leg stance, with no restrictions placed on upper-limb movement. The maximum measurement time was 60 s. Participants who were unable to maintain this standing position for 60 s at the first attempt were tested a second time. The SLS test has been shown to be a valid and reliable measure of balance in older adults.20,21)
To evaluate gait function, walking speed was measured at each participant’s preferred pace. Participants were instructed to walk along a 5-m path, with an additional 1 m section provided at each end for acceleration and deceleration. The walk was performed once, and the time was recorded in seconds.
The two-step test, the stand-up test, and the GLFS-25 were used to assess the risk and severity of LS.22) The two-step test evaluates horizontal mobility function.23) Starting from a standing position, participants took two maximal strides forward while maintaining balance. The test was performed twice, and the better result was recorded. The distance covered was standardized by dividing it by the participant’s height.
The stand-up test assesses vertical mobility function.24) This test is performed using stools of four different heights (10, 20, 30, and 40 cm). Participants were asked to stand up from each stool using one or both legs. A trial was considered successful if the participant maintained the standing position for more than 3 s without additional steps. Performance was graded according to the standard diagnostic criteria established by JOA.
The GLFS-25 is a self-administered questionnaire that assesses pain, physical function, and disability.25) Participants completed the 25-item GLFS-25, with each item scored from 0 to 4. The total score, ranging from 0 to 100, was used for analysis, with higher scores indicating more severe functional impairment.
Statistical MethodsThe hypothetical model proposed in this study assumes that skeletal muscle mass and PBF are associated with LS risk through balance and gait functions. Participants with missing data on key variables were excluded, and a complete case analysis was conducted. Accordingly, SEM was performed to evaluate the fit of the hypothesized model. To determine the required sample size for SEM, a power analysis based on the root mean square error of approximation (RMSEA) was conducted following the method proposed by MacCallum et al.26) The analysis was performed using the semPower package in R. The parameters for the analysis were set as follows: α = 0.05, desired power = 0.80, degrees of freedom = 10, RMSEA under the null hypothesis = 0.08 (indicating poor fit), and RMSEA under the alternative hypothesis = 0.05 (representing good fit). The analysis showed that a minimum sample size of approximately 230 participants was required to detect a difference between these two levels of model fit with sufficient statistical power.
The model’s goodness of fit was evaluated using the following statistical indices: goodness-of-fit index (GFI) >0.95, adjusted goodness-of-fit index (AGFI) >0.95, comparative fit index (CFI) >0.95, Tucker–Lewis index (TLI) >0.95, normed-fit index (NFI) >0.95, and RMSEA <0.07.27) Sensitivity analyses were conducted to examine the robustness of the associations. In these models, age and body mass index (BMI) were included as exogenous covariates with direct paths to skeletal muscle mass, PBF, balance, gait, and LS. Grip strength was also incorporated as a parallel pathway with direct paths to balance, gait, and LS. All sensitivity models were estimated using the same bootstrap procedure (5000 resamples) as used in the primary analysis. All statistical analyses were performed using SPSS Statistics version 23.0 (IBM, Armonk, NY, USA) and AMOS version 23.0.0.
A total of 298 women were initially recruited. Among them, 6 (2.0%) had missing values on LS tests (two-step test, stand-up test, or GLFS-25), and 7 (2.3%) had missing values on body composition measures (SMI or PBF). Because the overall proportion of missing data was low and no systematic pattern was observed, the data were considered missing completely at random. After excluding 13 participants with missing data, 285 community-dwelling older women were included in the final analysis (Table 1). Table 2 summarizes the distributions of observed variables. Notably, SLS time showed a skewed distribution with a ceiling at 60 s (median = 40.2 s; 90th percentile = 60 s), indicating a potential ceiling effect.
| Variable | Value |
| Age, years | 74.1 ± 5.4 |
| Height, cm | 152.7 ± 5.8 |
| Weight, kg | 52.6 ± 8.0 |
| BMI, kg/m2 | 22.5 ± 3.0 |
| SMI, kg/m2 | 5.7 ± 0.6 |
| PBF, % | 32.3 ± 6.6 |
| Hand grip, kg | 22.1 ± 3.7 |
| SLS time, s | 37.0 ± 22.4 |
| Walking speed, m/s | 1.5 ± 0.2 |
| Two-step test | 1.39 ± 0.15 |
| Stand-up test | |
| Grade 0 | 116 |
| Grade 1 | 147 |
| Grade 2 | 14 |
| Grade 3 | 8 |
| GLFS-25 | 8.5 ± 8.8 |
Values are presented as mean ± standard deviation or as number.
| Variable | Minimum | 25th percentile | Median | 75th percentile | 90th percentile | 95th percentile | Maximum |
| SMI, kg/m2 | 3.9 | 5.2 | 5.6 | 6.1 | 6.5 | 7 | 7.6 |
| PBF, % | 11.9 | 28.1 | 32.4 | 36.9 | 40.6 | 45 | 49.5 |
| SLS, s | 0 | 14.8 | 40.2 | 60 | 60 | 60 | 60 |
| Walking speed, m/s | 0.77 | 1.35 | 1.47 | 1.55 | 1.78 | 2.02 | 2.33 |
| Stand-up test | 0 | 1 | 1 | 2 | 2 | 3 | 3 |
| Two-step test | 0.82 | 1.19 | 1.29 | 1.39 | 1.55 | 1.68 | 1.9 |
| GLFS-25 | 0 | 3 | 6 | 12 | 18 | 33.7 | 61 |
The SEM demonstrated marginal overall fit: χ2(10) = 37.68, P < 0.001; GFI = 0.963; AGFI = 0.897; CFI = 0.927; TLI = 0.847; NFI = 0.906; RMSEA = 0.099 [90% confidence interval (CI): 0.066–0.133]. The standardized factor loadings of the LS latent construct were 0.67 for the stand-up test, −0.76 for the two-step test, and 0.54 for the GLFS-25, indicating that higher latent scores reflect worse locomotive function. The model explained 15% of the variance in walking speed, 6% in balance function (SLS time), and 54% in LS risk (latent variable). Although indices such as GFI, CFI, and NFI indicated acceptable fit, the elevated RMSEA and suboptimal TLI suggested potential model misfit (Fig. 1A).
Path analysis of the associations between body composition and LS risk via physical function using a cross-sectional structural equation model. (A) Initial hypothetical model, illustrating how SMI and PBF may be associated with LS risk via walking speed and SLS time. (B) Revised model that additionally includes a direct association between SMI and LS risk. Each model includes SMI and PBF as predictors of walking speed and SLS time, which in turn are associated with LS risk. LS risk was further assessed using the stand-up test, two-step test, and the GLFS-25. Solid arrows, statistically significant associations; dashed arrows, non-significant associations (N.S.); numerical values, standardized path coefficients.
SMI; Skeletal Muscle Index, PBF; Percent Body Fat, SLS; Single-Leg Standing, LS; Locomotive Syndrome; GLFS-25; 25-question Geriatric Locomotive Function Scale.
The initial model was then modified based on the modification indices (MI). The indices suggested additional paths from SMI to LS (MI = 11.039), from SMI to the stand-up test (MI = 18.054), and from PBF to the stand-up test (MI = 5.387). These paths were incorporated into a revised model, and the goodness of fit was recalculated (Fig. 1B). Consequently, the final revised model demonstrates acceptable fit: χ2(7) = 13.226, P = 0.067; GFI = 0.986; AGFI = 0.945; CFI = 0.984; TLI = 0.951; NFI = 0.967; RMSEA = 0.056 (90% CI: 0.000–0.102). The chi-square test was not significant, and the RMSEA 90% CI fell within the acceptable range. Other indices, including GFI, CFI, and TLI, also indicated acceptable fit. Table 3 presents the full model fit results. Because the revised model was derived through post-hoc modifications based on modification indices, this model should be interpreted as exploratory rather than confirmatory. Collectively, these results suggest that the revised structural model adequately fits the data, given the sample size (n = 285). The standardized factor loadings of the LS latent construct were 0.65 for the stand-up test, −0.76 for the two-step test, and 0.54 for the GLFS-25, indicating that higher latent scores represented worse locomotive function. The model explained 15% of the variance in walking speed, 6% in balance function (SLS time), and 56% in LS risk (latent variable).
| Index | Hypothetical model | Revised model |
| GFI | 0.963 | 0.986 |
| AGFI | 0.897 | 0.945 |
| CFI | 0.927 | 0.984 |
| TLI | 0.847 | 0.951 |
| NFI | 0.906 | 0.967 |
| RMSEA (90% CI) | 0.099 (0.066, 0.133) | 0.056 (0.000, 0.102) |
In the revised model, the path coefficients from PBF to balance function (β = −0.24) and from balance function (SLS time) to LS risk (β = −0.37) were significant, whereas the path from skeletal muscle mass to balance function was not significant. Bootstrap analyses (5000 resamples) showed that PBF had a significant positive indirect association with LS (β = 0.186, 95% CI = 0.092–0.280, P <0.01), whereas skeletal muscle mass exhibited a small negative indirect association (β = −0.097, 95% CI = −0.201 to −0.003, P <0.05) but no significant total effect (β = 0.010, 95% CI = −0.149 to 0.179). Tables 4–6 show the full results for direct, indirect, and total effects with 95% CIs, and the correlation matrix of observed variables.
| Index | Direct effect | Indirect effect | Total effect |
| Hypothetical model | |||
| SMI | 0.000 (0.000, 0.000) | −0.094 (−0.192, −0.003)* | −0.094 (−0.192, −0.003) |
| PBF | 0.000 (0.000, 0.000) | 0.184 (0.091, 0.276)** | 0.184 (0.091, 0.276)** |
| Revised model | |||
| SMI | 0.107 (−0.018, 0.237) | −0.097 (−0.201, −0.003)* | 0.010 (−0.149, 0.179) |
| PBF | 0.000 (0.000, 0.000) | 0.186 (0.092, 0.280)** | 0.186 (0.092, 0.280)** |
Bootstrap method with 5000 resamples, bias-corrected 95% confidence intervals shown in parentheses.
* P <0.05; ** P <0.01
| Variable | SLS time | Walking speed | LS risk | Stand-up test | Two-step test | GLFS-25 |
| Hypothetical model | ||||||
| SMI | ||||||
| Direct | 0.058 | 0.123 | 0 | 0 | 0 | 0 |
| Indirect | 0 | 0.019 | −0.094 | −0.063 | 0.072 | −0.051 |
| Total | 0.058 | 0.143 | −0.094 | −0.063 | 0.072 | −0.051 |
| PBF | ||||||
| Direct | −0.239 | −0.103 | 0 | 0 | 0 | 0 |
| Indirect | 0 | −0.080 | 0.184 | 0.122 | −0.140 | 0.099 |
| Total | −0.239 | −0.183 | 0.184 | 0.122 | −0.140 | 0.099 |
| SLS time | ||||||
| Direct | 0 | 0.333 | −0.383 | 0 | 0 | 0 |
| Indirect | 0 | 0 | −0.168 | −0.366 | 0.421 | −0.298 |
| Total | 0 | 0.333 | −0.551 | −0.366 | 0.421 | −0.298 |
| Walking speed | ||||||
| Direct | 0 | 0 | −0.504 | 0 | 0 | 0 |
| Indirect | 0 | 0 | 0 | −0.336 | 0.385 | −0.273 |
| Total | 0 | 0 | −0.504 | −0.336 | 0.385 | −0.273 |
| LS risk | ||||||
| Direct | 0 | 0 | 0 | 0.665 | –0.764 | 0.542 |
| Indirect | 0 | 0 | 0 | 0 | 0 | 0 |
| Total | 0 | 0 | 0 | 0.665 | −0.764 | 0.542 |
| Revised model | ||||||
| SMI | ||||||
| Direct | 0.058 | 0.123 | 0.107 | 0.154 | 0 | 0 |
| Indirect | 0 | 0.019 | −0.097 | 0.006 | −0.007 | 0.005 |
| Total | 0.058 | 0.143 | 0.010 | 0.160 | −0.007 | 0.005 |
| PBF | ||||||
| Direct | −0.239 | −0.103 | 0 | 0.078 | 0 | 0 |
| Indirect | 0 | −0.080 | 0.186 | 0.120 | −0.141 | 0.101 |
| Total | −0.239 | −0.183 | 0.186 | 0.198 | −0.141 | 0.101 |
| SLS time | ||||||
| Direct | 0 | 0.333 | −0.372 | 0 | 0 | 0 |
| Indirect | 0 | 0 | −0.176 | −0.355 | 0.415 | −0.298 |
| Total | 0 | 0.333 | −0.548 | −0.355 | 0.415 | −0.298 |
| Walking speed | ||||||
| Direct | 0 | 0 | −0.529 | 0 | 0 | 0 |
| Indirect | 0 | 0 | 0 | −0.343 | 0.401 | −0.288 |
| Total | 0 | 0 | −0.529 | −0.343 | 0.401 | −0.288 |
| LS risk | ||||||
| Direct | 0 | 0 | 0 | 0.648 | −0.758 | 0.544 |
| Indirect | 0 | 0 | 0 | 0 | 0 | 0 |
| Total | 0 | 0 | 0.648 | −0.758 | 0.544 | |
| Variable | SMI | PBF | SLS time | Walking speed | Stand-up test | Two-step test |
| PBF | 0.200 | |||||
| SLS time | 0.010 | −0.227 | ||||
| Walking speed | 0.106 | −0.154 | 0.358 | |||
| Stand-up test | −0.038 | 0.110 | −0.375 | −0.427 | ||
| Two-step test | 0.044 | −0.126 | 0.430 | 0.490 | −0.508 | |
| GLFS-25 | −0.031 | 0.089 | −0.305 | −0.347 | 0.361 | −0.414 |
Sensitivity analyses, including age or BMI as covariates, attenuated model fit but did not change the main findings: PBF retained a significant positive indirect association with LS, and SMI showed a small negative indirect association. Similarly, when grip strength was included as a parallel pathway, the pattern of associations remained consistent. Tables 7 and 8 present the complete results of these analyses.
| Variable/index | Age, BMI |
| Indirect effect on LS | |
| SMI (95% CI) | −0.088 (−0.182, −0.004)* |
| PBF (95% CI) | 0.165 (0.081, 0.251)** |
| Fit indices | |
| GFI | 0.943 |
| AGFI | 0.828 |
| CFI | 0.940 |
| TLI | 0.857 |
| NFI | 0.929 |
| RMSEA (90% CI) | 0.125 (0.099, 0.152) |
Bootstrap method with 5000 resamples, bias-corrected 95% confidence interval.
* P <0.05; ** P <0.01.
| Variable/index | Hypothetical model | Revised model |
| Indirect effect on LS | ||
| SMI (95% CI) | −0.067 (−0.181, 0.046) | −0.157 (−0.286, −0.039)** |
| PBF (95% CI) | 0.107 (0.015, 0.200)* | 0.099 (0.011, 0.188)* |
| Fit indices | ||
| GFI | 0.944 | 0.968 |
| AGFI | 0.845 | 0.871 |
| CFI | 0.893 | 0.944 |
| TLI | 0.770 | 0.824 |
| NFI | 0.875 | 0.930 |
| RMSEA (90% CI) | 0.123 (0.096, 0.153) | 0.108 (0.074, 0.144) |
Bootstrap method with 5000 resamples, bias-corrected 95% confidence interval.
* P <0.05; ** P <0.01.
Although the final model did not meet all the stringent cut-off values recommended by Hooper et al.,27) (specifically AGFI, TLI, and NFI >0.95 and RMSEA <0.07), it did satisfy widely accepted thresholds for acceptable model fit. According to Hu and Bentler,28) model fit is considered acceptable when CFI and TLI are 0.90 or greater and RMSEA is 0.08 or lower. In this study’s model, CFI = 0.984, TLI = 0.951, and RMSEA = 0.056, which collectively support the conclusion that the model provides an acceptable representation of the data, despite falling short of more conservative criteria.
Low skeletal muscle mass has been considered a key target for interventions in individuals with LS. However, this study found that higher PBF was significantly associated with greater LS risk, partly explained by differences in balance (SLS time) and walking speed, rather than skeletal muscle mass. These findings suggest that, for older women with LS, maintaining an appropriate balance between skeletal muscle mass and PBF may be important for better physical function and lower LS risk.
The novelty of this study lies in highlighting that high PBF is significantly associated with increased LS risk, with this relationship statistically linked to physical function in older women. Although previous studies have independently linked sarcopenia, PBF, balance, and gait functions to LS,10,11,12,14,16) the interactions among these combined risk factors remain unclear. In this study, we observed associations consistent with a mediation hypothesis, indicating that the combined presence of these factors is associated with greater LS risk. A key strength of this study is the identification of SLS time and walking speed as statistical indicators consistent with the hypothesized associations among balance, gait, and LS risk. However, these indirect associations were modest in magnitude and should be interpreted as exploratory, given the cross-sectional design and potential unmeasured confounding. These findings may inform approaches for LS prevention research.
For community-dwelling older women, both low skeletal muscle mass and high PBF should be considered in strategies to reduce LS risk. Previous studies have demonstrated that high PBF is associated with mobility-related disability in older adults.29) Additionally, women generally have lower lean body mass than men,2) making high PBF a biomechanical disadvantage that can contribute to balance dysfunction.4) Consequently, older women with higher PBF may have an increased association with balance dysfunction, gait impairment, and LS risk. Notably, sensitivity analyses including BMI as a covariate confirmed that these associations remained significant, indicating that the influence of adiposity did not materially alter the main findings.
As a preventive exercise for LS, the JOA recommends locomotion training (LT), which is a safe, simple, and effective multifactorial program that includes balance and muscle strengthening.17) The primary aim of LT is to strengthen the lower-limb muscles and improve balance, which are essential for walking and performing basic activities of daily living. This study showed that higher PBF was associated with lower balance function, which was statistically related to slower walking speed and greater LS risk. Therefore, for older women with high PBF, LT exercises that focus on improving balance, such as SLS, may be particularly effective in reducing LS risk by enhancing walking speed.30) In addition to balance-focused exercises such as LT, targeted interventions for individuals with high PBF should also be considered. Dietary modifications (e.g., balanced caloric restriction, increased protein intake, and reduced consumption of pro-inflammatory foods) may help lower body fat and improve physical function.31,32,33) Aerobic exercise can reduce body fat while enhancing balance and walking ability.34,35,36) Furthermore, lifestyle-based strategies, including increased daily physical activity and reduced sedentary behavior, may mitigate LS risk.37) Integrating these approaches with existing LS prevention strategies could enhance their overall effectiveness. Future longitudinal and intervention studies are warranted to evaluate the combined impact of dietary, aerobic, and lifestyle-based interventions on LS prevention in older women.
Although the direct path from SMI to LS risk was not statistically significant in the revised model, model fit indices improved slightly when this path was retained, suggesting that retaining this pathway statistically captured some unexplained variance in the observed data. However, this post-hoc modification was exploratory and should be interpreted with caution. The positive association observed between SMI and the stand-up test performance (β = 0.15) is likely a statistical artifact rather than a genuine clinical association. This path may reflect residual covariance between SMI and PBF, or limited variance in SMI within this relatively healthy cohort, rather than a true detrimental effect of higher muscle mass. Therefore, the final model should be regarded as an empirical representation of the observed associations rather than a confirmatory model of physiological mechanisms. Future studies using larger and more heterogeneous samples, as well as longitudinal designs, are warranted to validate and refine these exploratory associations.
The study has several limitations. First, all participants were women, preventing analysis of sex differences, which may involve different mechanisms for LS onset and should be investigated in future studies. Second, the cross-sectional design precludes the determination of causal associations with LS risk. Longitudinal studies are necessary to identify causal factors and guide the development of targeted interventions to prevent LS onset. Third, balance function was assessed using only the SLS test. Although SLS is a simple and widely used measure, balance is a multifaceted construct involving muscle strength, vestibular function, and cognitive processing. Therefore, reliance on SLS alone may not fully capture overall balance performance. Additionally, the distribution of SLS time was skewed, with a ceiling at 60 s, indicating a potential ceiling effect. Future studies should consider using more comprehensive assessment tools, such as the Berg Balance Scale, and statistical methods that account for censoring (e.g., Tobit models) to provide a more robust evaluation of balance function.38) Fourth, another limitation relates to the SEM approach. By design, SEM can only account for variables explicitly included in the model. In our analysis, physical activity level—which may influence both PBF and LS risk—was not measured.39,40,41) Physical activity could act as a confounder (e.g., individuals with higher activity levels tend to have lower PBF and lower LS risk) or as a mediator (e.g., higher adiposity leading to reduced activity, which in turn increases LS risk). The omission of this variable suggests that part of the observed associations may be attributable to unmeasured effects of physical activity. Future studies should incorporate physical activity measures to better delineate the causal pathways linking body composition, physical function, and LS risk. Fifth, participants were recruited from sports and regional community centers and were relatively active and healthy older Japanese women, which may introduce a healthy-volunteer bias. Information on medical history, comorbidities, and lifestyle factors such as smoking and alcohol consumption was not collected, further limiting the characterization of the study population. The physical function and body composition values indicated that most participants did not meet the criteria for sarcopenia, and many were classified as LS Stage 0 or 1. Therefore, the external validity of these findings is limited, and they should be interpreted as reflecting associations observed primarily in the early stages of LS. Caution is warranted when generalizing to less active populations, individuals with comorbidities, or patients with advanced LS. Future studies involving more heterogeneous and frail populations are needed to validate and extend these findings. Furthermore, the SEM used in this study specified linear associations among variables, and potential non-linear dose–response relationships (e.g., U-shaped patterns) were not examined. Larger and more heterogeneous samples are required in future studies to explore such possibilities.
In a cohort of community-dwelling older women, higher PBF was associated with lower balance and gait functions, which were statistically related to greater LS risk. These observed associations, found in predominantly non-sarcopenic older women, may inform future studies and preventive approaches targeting PBF.
The authors thank the staff members of the participating health centers and the participants for their contributions to the study. This study was supported by the Ministry of Health, Labour, and Welfare of Japan (Grant Number: 22FA1003).
The authors declare no conflict of interest.
