INTRODUCTION
Vertebral compression fractures (VCFs) are common among older adults and are often precipitated by age-related declines in physical function and falls.1) VCFs are commonly associated with severe pain and spinal deformity, which in turn lead to reduced walking ability and limited activities of daily living (ADL).2) Therefore, in patients with VCFs, the assessment and improvement of gait function represent important clinical issues for achieving post-discharge mobility independence and maintaining ADL.
One of the primary approaches to supporting gait function in older adults is the use of walking aids. Cane use has been reported to alleviate pain and enhance activity levels and social participation,3) whereas walker use improves gait function by increasing the base of support.4) In addition, gait modality at discharge, such as independent ambulation, cane-assisted walking, or walker-assisted walking, has been associated with the risk of future readmission.5) Accordingly, identifying factors associated with gait modality is essential for preventing functional decline. However, inappropriate use of walking aids may increase the risk of falls,6,7) underscoring the importance of selecting appropriate assistive devices.
Several studies have investigated the use of walking aids. For example, in patients with hip fractures, factors such as age and cognitive function are associated with the type of walking aid used.8) Among community-dwelling older adults, balance ability and depressive symptoms have been associated with cane use.9) Although greater declines in physical function are generally associated with the use of more supportive walking aids, it remains unclear which specific domains of physical function are most strongly associated with walking aid use at discharge in patients with VCFs. Importantly, previous studies have largely focused on other populations, and evidence specific to VCF patients—who often present with pain-related functional limitations and spinal deformity—is limited.
Furthermore, walking aid prescription and functional status are typically assessed at the same time point at discharge, making it difficult to distinguish whether functional limitations lead to the selection of walking aids or whether walking aid use reflects underlying clinical decisions. Therefore, these relationships should be interpreted as associations rather than causal effects. Accordingly, the purpose of this study was to examine the associations between physical function, pain, and walking aid use at discharge in patients with VCFs. By focusing on discharge as a key clinical decision-making point, this study aims to provide insights into which functional domains are most closely related to walking aid selection, thereby informing rehabilitation practice and discharge planning.
MATERIALS AND METHODS
In this retrospective cross-sectional study, patients diagnosed with VCFs based on the Denis classification10,11) were included. The Denis classification is a widely used system for assessing fracture stability and clinical severity, as it categorizes injury location based on the three-column structure of the spine. In this study, VCFs were defined as fractures involving compression of the anterior vertebral elements, with injury confined to the anterior supporting structures, thereby ensuring a more homogeneous study population.
To ensure diagnostic accuracy, all patients were evaluated using plain radiography and magnetic resonance imaging according to the standard diagnostic protocol at our institution. Diagnoses were made by orthopedic surgeons and rehabilitation physicians and, when necessary, confirmed by multiple physicians. Patients aged ≥65 years with VCFs who were discharged from the convalescent rehabilitation ward of Kurume Rehabilitation Hospital, Kabutoyama-kai Medical Corporation, between August 2019 and June 2024 were included.
Patients were excluded if they had used a wheelchair before admission; were admitted after undergoing surgery in an acute care hospital; had difficulty understanding instructions; were transferred to another facility or ward or died during hospitalization; had their diagnosis revised due to changes in clinical condition; had pacemaker implantation; became unable to walk during hospitalization due to poor physical condition; or were suspected of having severe cognitive impairment (Hasegawa’s Dementia Scale–Revised [HDS-R] <13).12)
Gait status at discharge was classified into two groups based on in-hospital walking ability: the independent ambulation group (patients who could walk without assistive devices) and the walking aid group (patients who used a cane or walker). Walking aids at discharge were selected by the attending rehabilitation professionals (physical and/or occupational therapists) based on comprehensive clinical assessments of physical function, including balance, muscle strength, pain, and overall mobility. No standardized protocol was applied; decisions were made according to routine clinical judgment.
This study was approved by the Ethics Committee of Kurume Rehabilitation Hospital, Kabutoyama-kai Medical Corporation (Approval No. 26-001). Data were retrospectively extracted and analyzed from medical records with strict protection of personal information. Comprehensive consent for the use of medical records was obtained from all patients at admission, and participation was based on an opt-out approach due to the retrospective design.
Assessments
The variables examined included age, sex, pain intensity, HDS-R, handgrip strength, 30-second chair stand test (CS-30), Berg Balance Scale (BBS), skeletal muscle mass index (SMI), and in-hospital gait modality at discharge.
Pain intensity was assessed using an 11-point numerical rating scale (NRS),13) ranging from 0 (“no pain”) to 10 (“worst imaginable pain”). In this study, pain during walking in daily life at discharge was used as the representative measure. Pain assessment was not restricted to a standardized time point or specific activity but reflected pain experienced during walking at the time of clinical evaluation. When multiple assessments were available, the value closest to discharge was used for analysis.
The HDS-R is a brief screening tool for cognitive function consisting of nine items with a maximum score of 30,12) and the score at discharge was used for analysis.
The CS-30 is a reliable and valid measure of lower extremity muscle strength.14) Assessments were conducted using a chair without armrests and with a seat height of 40 cm, with careful consideration of pain, and the value at discharge was used.
The BBS is an established measure for assessing dynamic and static balance15) and is associated with fall risk and gait function in patients with VCFs16) and hip fractures.17) The BBS score ranges from 0 to 56, and assessments were conducted with sufficient attention to pain and fall prevention, with the score at discharge used for analysis.
Handgrip strength18) was measured at discharge using a digital dynamometer (TKK5401; Takei Scientific Instruments Co. Ltd., Niigata, Japan). Measurements were performed in the seated position with the elbow in full extension, and grip width was adjusted so that the proximal interphalangeal joint of the index finger was positioned at approximately 90° during measurement. Each hand was assessed twice, and the maximum value obtained from either side was used for analysis to account for potential asymmetry due to pain or functional impairment. This measurement position was adopted to ensure patient safety and minimize discomfort due to residual pain.
SMI was measured using a body composition analyzer (InBody S10; InBody Co. Ltd., Seoul, Korea). Measurements were performed in the supine position after a brief rest period. Muscle mass of the upper limbs, lower limbs, and trunk was assessed, and SMI was calculated by dividing the sum of appendicular muscle mass by height squared (kg/m²).19) However, strict control of hydration status was not implemented, as measurements were conducted under routine clinical conditions.
Walking ability at discharge was categorized into independent ambulation within the hospital and ambulation with the use of assistive devices (cane or walker). The mode of ambulation was determined by the attending physical and occupational therapists based on comprehensive clinical evaluations, including balance, muscle strength, and pain.
Statistical Analysis
For the statistical analysis, participants were first classified into the independent ambulation group and the walking aid group (cane or walker) based on in-hospital gait status at discharge. As the primary analysis, these groups were compared. In addition, to examine differences in gait modality within the walking aid group, an exploratory analysis was conducted comparing the cane-assisted walking group and the walker-assisted walking group. For between-group comparisons, the Mann–Whitney U test was used for continuous variables, and the chi-square test for nominal variables. Effect sizes for continuous variables were calculated as r and those for nominal variables as the phi (φ) coefficient, with magnitudes interpreted as small (≥0.1), medium (≥0.3), and large (≥0.5).20) Effect sizes were expressed as absolute values. In the primary analysis, a binomial logistic regression analysis was conducted with the independent ambulation group as the reference category. Independent variables included pain at discharge, HDS-R, handgrip strength, CS-30, BBS, and SMI, with age and sex as covariates. Furthermore, an exploratory binomial logistic regression analysis was conducted within the walking aid group using the cane-assisted walking group as the reference. To assess the robustness of the model estimates, a bootstrap resampling procedure with 1,000 iterations was applied. Multicollinearity was assessed using the variance inflation factor. Additionally, receiver operating characteristic (ROC) curve analysis was conducted to determine cutoff values for discriminating between independent ambulation and walking aid use in the primary analysis. Discriminative ability was interpreted based on area under the curve (AUC) values as follows: 0.70–0.79 acceptable, 0.80–0.89 good, and ≥0.90 excellent.21,22) Statistical significance was set at p<0.05. All analyses were performed using SPSS Statistics version 28.0 (IBM Corp., Armonk, NY, USA).
RESULTS
A total of 143 patients were included in the final analysis (Fig. 1). The cohort comprised 28 males and 115 females, with a median age of 86.0 years (interquartile range, 81.0–90.0). Regarding fracture characteristics, 124 (87%) patients had single-level fractures, and 19 (13%) had multiple-level fractures. Fracture locations were the thoracic spine in 56 (39%) patients, lumbar spine in 74 (52%), and thoracolumbar junction in 13 (9%). The mechanisms of injury included falls in 63 patients (44%), ADL or leisure activities in 68 (48%), and unknown causes in 12 (8%). Additionally, 79 (55%) patients had osteoporosis. Based on in-hospital gait status at discharge, patients were classified into the independent ambulation group (n=54) and the walking aid group (n=89). The walking aid group was further subdivided into the cane group (n=50) and the walker group (n=39). Among patients using walkers, 15 used standard walkers and 24 used forearm support walkers.
In the primary analysis comparing the independent ambulation and walking aid groups, significant differences were observed in age, pain, handgrip strength, CS-30, BBS, and SMI (all p<0.05). Regarding effect sizes, handgrip strength and CS-30 showed moderate effects, whereas BBS demonstrated a large effect (Table 1). Additionally, within the walking aid group, comparisons between the cane-assisted and walker-assisted groups revealed significant differences in pain, handgrip strength, CS-30, BBS, and SMI (all p<0.05). Small effects were observed for handgrip strength and SMI, moderate effects for pain and CS-30, and a large effect for BBS (Table 2).
In the binomial logistic regression analysis with independent ambulation as the reference category, BBS at discharge was significantly associated with walking aid use (odds ratio [OR] per 1-point increase in BBS, 0.85; 95% confidence interval [CI] 0.77–0.93; p<0.001). Briefly, higher BBS scores were associated with lower odds of walking aid use. A bootstrap resampling procedure (1,000 iterations) confirmed the robustness of these results. Model fit was supported by a significant likelihood ratio test (χ²=62.8, degrees of freedom=8, p<0.001) and a nonsignificant Hosmer–Lemeshow test (p=0.20).
In the exploratory analysis using binomial logistic regression, with the cane group as the reference category, pain at discharge and SMI were significantly associated with walker use—pain (OR per 1-unit increase in NRS=1.43; 95% CI 1.03–1.96; p=0.029) and SMI (OR per 1-unit increase=0.40; 95% CI 0.20–0.79; p=0.010). These findings were also supported by the bootstrap analysis. Model fit was supported by a significant likelihood ratio test (χ²=28.66, degrees of freedom=8, p<0.001) and a nonsignificant Hosmer–Lemeshow test (p=0.76) (Table 3).
ROC curve analysis revealed that BBS at discharge had good discriminative ability for distinguishing between the independent ambulation and walking aid groups (AUC=0.83; 95% CI 0.76–0.90; p<0.001). The optimal BBS cutoff was 48.5 points, with a sensitivity of 79.2% and specificity of 75.0% (Table 4, Fig. 2).
DISCUSSION
In this retrospective cross-sectional study, BBS at discharge was identified as the principal factor associated with walking aid use in patients with VCFs. In addition, exploratory analyses suggested that pain and SMI may be associated with differentiating cane-assisted from walker-assisted walking.
The association between BBS and gait status underscores the importance of balance and postural control in determining walking aid use. BBS is widely used to assess fall risk and functional independence and reflects multiple components of functional stability required for safe ambulation.9,23) In the present study, BBS also distinguished independent ambulation from walking aid use. In contrast, CS-30 was not independently associated with gait status in the multivariable analysis, although associations were suggested in the univariate and bootstrap analyses. Therefore, CS-30 may be associated with walking aid use; however, this finding should be interpreted with caution given the limited sample size and potential instability of the estimates.
In the exploratory analysis, both pain and SMI were associated with differentiating cane-assisted from walker-assisted walking. Reduced SMI may reflect decreased muscle mass and sarcopenia, which are associated with impaired physical function and increased fall risk.24-26) Furthermore, reduced SMI may correspond to criteria proposed by the Asian Working Group for Sarcopenia,24) suggesting that sarcopenia-related declines in physical function may partly underlie walking aid selection. Therefore, patients with lower SMI may be more likely to require more stable walking aids such as walkers. However, these findings should be interpreted with caution due to the exploratory nature of the analysis, the small sample size, and the potential influence of hydration status on bioelectrical impedance measurements.
Pain, a major symptom of VCFs, may also influence walking aid selection through gait instability and movement avoidance.27) In addition, although not assessed in this study, fear of falling may represent an important unmeasured factor. Increased pain or gait instability may heighten concerns about falling, potentially contributing to the selection of more stable walking aids.
In the ROC analysis of BBS, the AUC observed in this study was considered to indicate good discriminative ability based on commonly used interpretation criteria (≥0.70 acceptable, ≥0.80 good).21,22) However, because this study was based on a single-center dataset and a cross-sectional assessment at discharge without external validation, these findings, including the proposed cutoff, should be interpreted as exploratory, and further validation is required before clinical application. Previous studies have reported BBS thresholds of approximately 45 points for fall risk and gait independence in patients with VCFs,16,28) and the cutoff identified in this study is consistent with these findings, which may support its plausibility.
This study has several limitations. First, this cross-sectional design based on physical function at discharge does not account for changes in physical function or pain during hospitalization or recovery trajectories. Second, unmeasured confounding factors, such as trunk muscle strength, psychological status (e.g., fear of falling and depression), and analgesic use, may have influenced walking aid selection. Third, this study did not fully distinguish differences in functional demands among types of walkers. Fourth, walking aid use at discharge was determined based on clinician judgment without standardized criteria, and inter-rater variability cannot be excluded. Fifth, the relatively large number of explanatory variables relative to the sample size raises a risk of overfitting, despite the use of bootstrap methods. In addition, the small sample size, particularly in the walker group, may have contributed to instability in the multivariable estimates. Despite these limitations, this study provides additional insights into factors associated with the type of walking aid used at discharge in patients with VCFs, highlighting the clinical importance of selecting appropriate walking aids based on objective indicators.
Conclusion
In this study, walking aid use at discharge in patients with VCFs was associated with BBS. In addition, exploratory analysis indicated that walker use among patients using walking aids was associated with pain and SMI. These findings suggest the importance of assessing balance function, pain severity, and skeletal muscle mass when evaluating gait modality at discharge. However, as this was a single-center retrospective observational study, further studies, including external validation, are required to confirm the cutoff values and the findings from the exploratory analysis.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
