Principal Component Analysis to Examine Fall-predictive Screening Tests for use during Locomotive Health Checkups

ABSTRACT

Objectives: The purpose of this study was to determine which components of screening tests are important for assessing locomotive dysfunction during locomotive health checkups and best predict falls in a prospective cohort study.

Methods: Four hundred and sixty-eight residents were assessed for locomotive syndrome, frailty, knee and back pain, bone mass, muscle mass, grip strength, gait speed, gait variability, and kyphosis at a baseline locomotive health checkup. Residents were followed up after 1 year and surveyed about their incidence of falls.

Results: A total of 379 residents were analyzed in the study. Principal component analysis was used to divide components using all screening tests for locomotive dysfunction. The primary variable making up the first principal component was “mobility function,” the second was “muscle function,” and the third was related to “spinal alignment.” The tests that showed the highest principal component loadings in each component were the five-question Geriatric Locomotive Function Scale, muscle mass index, and kyphosis index. Binary regression analyses, adjusted for age and past fall history, showed that the five-question Geriatric Locomotive Function Scale was independently related with the incidence of falling (odds ratio=2.04; 95% confidence interval: 1.04–3.99).

Conclusions: We propose that during a locomotive health checkup, mobility function, muscle function, and spinal alignment are important components for the assessment of locomotive dysfunction. Notably, the five-question Geriatric Locomotive Function Scale is a simple and convenient screening test that can predict the future incidence falls.

INTRODUCTION

In Japan, the main causes of the need for nursing care are “dementia,” “cerebrovascular disease,” “frailty,” “fractures and falls,” and “joint diseases.” Among these, when frailty, fractures and falls, and joint diseases are combined to define “locomotive dysfunction,” they account for 37.3% of the total, making locomotive dysfunction the most common reason for nursing care.¹⁾ The qualitative deterioration and quantitative decline of locomotive organs as a result of osteoarthritis, osteoporosis, various joint diseases, and age-related sarcopenia can impair joint function, reduce mobility, and adversely affect balance, all of which can increase the likelihood of falls and fractures.²⁾ Therefore, for the prevention of fragility fractures in older individuals with osteopenia or osteoporosis for example, it is important to screen for locomotive dysfunction at an early stage.

Considering the aforementioned background, locomotive assessments were conducted during locomotive health checkups for community-dwelling older adults in hospitals—in collaboration with academic institutes and government agencies^3,4,5)—in an attempt to detect locomotive dysfunction at an early stage and offer appropriate treatment options and interventions. Various tests are available for the screening of locomotive dysfunction and/or assessment of physical function. The five-question Geriatric Locomotive Function Scale (GLFS-5) has been used to assess locomotive syndrome (LS),⁶⁾ and the Kihon Checklist (KCL)⁷⁾ has been employed to assess frailty and lower motor function in daily activity. Grip strength, muscle mass, and gait speed have been shown to have utility for the assessment of sarcopenia.⁸⁾ Quantitative ultrasound (QUS) is often used for measuring bone mass.^9,10) Gait variability is an objective test often used to assess gait stability and as an indicator of fall risk.^4,11) For the assessment of spinal function and/or kyphosis, the kyphosis index is typically used.¹²⁾ All of these modes of assessment have also been reported to be associated with the future occurrence of falls and fractures,^{9,13,14,15,16)} and the screening tests for locomotive dysfunction can be exceptionally useful in determining fall risk.

Although numerous tests can be used to screen for locomotive dysfunction, a number of them have overlapping components. For instance, it is well known that muscle mass and grip strength are typically well correlated. In addition, low bone mass assessed using QUS has been reported to correlate with low gait speed.¹⁷⁾ Rather than conducting a large battery of tests to screen for locomotive dysfunction among community residents, including those in long-term care settings, it is important to choose a test or tests that can most accurately assess the critical components that are more likely to discriminate fall risk. Conducting multiple assessments of locomotive dysfunctions and fall risk not only imposes a time burden on municipal care prevention staff but also complicates information sharing across various professions.

The purpose of this prospective cohort study was to identify the most effective component of locomotive dysfunction assessment in terms of its ability to predict the future incidence of falls. Principal component analysis was used to identify the screening tests with the highest principal component loadings as the strongest predictors of future falls.

MATERIALS AND METHODS

Subjects

This study requested participation from local residents who visited examination sites for specific health checkups and health checkups for older adults hosted by local government for 3 days every year. The following inclusion criteria were used: (1) aged older than 40 years (including those certified as “requiring assistance” in the long-term care insurance system), (2) living independently at their own home, and (3) able to walk without assistance. Those who had received “nursing care” in the long-term care insurance system were excluded. Study participants were newly included annually for 5 years from 2014 to 2018. In 2014, 223 new people participated. In addition, 99 participants were newly enrolled in 2015, 66 in 2016, 40 in 2017, and 40 in 2018. Overall, 468 participants were enrolled and received the baseline assessment in this study. Participants received follow-up assessments for 1 year after their first participation as a baseline and were assessed for the incidence of falling; that is, residents who newly participated in 2014 were followed up through 2015, and residents who participated in the study starting in 2018 were followed up through 2019.

This research was approved by the Institutional Review Board of the Kawasaki University of Medical Welfare (No: 21-109), and we conducted a secondary analysis of a previously used research database (GAINA study). We provided participants with the opportunity to opt out of the study.

Baseline Assessment

At the time of the town-sponsored medical checkup, the following baseline characteristics were recorded: age, gender, height, body weight, and body mass index (BMI). Participants were also assessed for medication use, eye and hearing problems, as well as any diagnoses of hypertension, diabetes, cardiovascular diseases, cerebrovascular disease, a previous diagnosis/history of osteoarthritis of the knee, osteoarthritis of the hip, lumbar disease, osteoporosis, and history of falls during the previous year.

Screening Tests

Locomotive Syndrome

LS was assessed using the GLFS-5,⁶⁾ a short version of the GLFS-25. The GLFS-5 is used to identify older adults at risk of developing increased dependency on others and/or adults requiring nursing care because of LS associated with musculoskeletal disorders. The GLFS-5 is self-administered but is relatively comprehensive. It consists of five items graded according to a 5-point scale ranging from no impairment (0 points) to severe impairment (4 points): item 1, “extent of difficulty with stairs;” item 2, “extent of difficulty to walk briskly;” item 3, “ability to walk without rest;” item 4, “extent of difficulty to carry objects;” item 5, “extent of difficulty performing load-bearing tasks and housework.” Each question is given a score out of 4 to give a total score out of 20. Low scores represent little or no impairment, whereas high scores indicate severe impairment. In the present study, a score greater than 6 was considered to indicate a diagnosis of LS.⁶⁾

Frailty

Frailty was defined based on the KCL, which was developed by the Japanese Ministry of Health, Labor, and Welfare. It is widely used for frailty screening in Japan.⁷⁾ The KCL is a simple yes/no questionnaire that can assess multiple aspects of motor function in daily living, such as instrumental activities of daily living, social living, motor, oral, and cognitive function, nutritional state, homebound status, and mood; responding ‘yes’ to any item counts as a point. The score range is 0 to 25 and higher scores can indicate frailty. In the present study, a KCL score of 8 or higher was considered to indicate frailty.^7,18)

Knee and Back Pain

Leg and back pain were assessed using a visual analog scale (VAS) score. Participants were asked if they had pain in their legs or back or both, and responses about any one of them were marked on a 10-cm line representing a continuum for the pain intensity (“no pain” at 0 cm to the “most severe pain” at 10 cm) experienced during the past few days.

Bone Mass

We used QUS to assess calcaneal bone mass. The speed of sound (SOS) through the calcaneus was measured using a CM-200 sonometer (Furuno, Nishinomiya, Japan). The subject was seated and placed their right heel, to which ultrasound gel was applied, on the QUS device. Two transducers were placed laterally on each side of the subject’s heel. The sound waves were then transmitted from one transducer through the calcaneus and were received by the other transducer. The % young adult mean (%YAM) was calculated from the SOS results.

Muscle Mass Index

Muscle mass was measured using bioelectrical impedance analysis (BIA) and a multi-frequency MC-780A segmental body composition analyzer (Tanita, Tokyo, Japan). Subjects stepped onto the analyzer and remained standing for approximately 30 s. Muscle mass was assessed as the muscle mass index, which was calculated by dividing limb muscle mass (kg) by height (m) squared. We considered that the muscle mass index indicated low muscle mass in men at values less than 7.0 kg/m² and less than 5.7 kg/m² in women.⁸⁾

Grip Strength

Muscle strength was assessed as grip strength, which was measured using a TKK-5401 dynamometer (Takei, Sanjo, Japan). The subject squeezed the dynamometer twice with each hand. The highest scores for the left and right hands were recorded as the representative values.

Gait Speed

Gait speed was assessed using an OptoGait analysis system (Microgate, Bolzano, Italy). The subjects completed a single trial after being instructed to “walk at your normal speed” across the middle 5 m of the test (i.e., between the 2-m and 7-m marks). The first and last 2 m were used to allow periods of acceleration and deceleration.

Gait Variability

Gait variability was assessed using a small (45 × 45 × 18.5 mm) tri-axial accelerometer (MVP-RF8, Micro Stone, Nagano, Japan). The accelerometer was attached to the third lumbar vertebra by a trunk belt that did not restrict the participant’s movements. The trunk belt was fixed firmly to the subject’s body to ensure motion artifacts were eliminated. Linear accelerations of the trunk were measured along the vertical axes at 200 Hz. All acceleration signals were synchronized and analyzed using specialized software (Wireless Vibration Recorder, MicroStone, Nagano, Japan). We analyzed the vertical axis accelerations using autocorrelation coefficients (ACs) to assess the severity of gait variability. ACs represent the correlation coefficients when the acceleration signal is shifted by the mean step or stride time period from the original signal.^4,19) AC values close to 1 indicate good regularity of gait. The ACs were calculated using SPSS statistical software (Version 22 for Windows; IBM, Tokyo, Japan).

Kyphosis Index

The kyphosis index was used to assess spinal alignment. After the examiner affixed orange ping-pong balls (as reference points) with adhesive tape to the participant’s clothed body (on the seventh cervical vertebra, the most prominent part of the spinal vertebra, and the fourth lumbar vertebra), the same examiner photographed the participant in the sagittal plane while they were standing with their arms crossed in front of their chest. Another examiner imported the photographs into image analysis software (ImageJ; National Institutes of Health, Bethesda, MD, USA), and the seventh cervical and the fourth lumbar spinous processes in the sagittal plane were connected by straight lines, X and Y, defined as the distance from the apex of the spinal curvature to the straight line. The kyphosis index was calculated as Y/X × 100.²⁰⁾ In the present study, a kyphosis index of 13% or more was considered the indicate the presence of kyphosis.²¹⁾

Follow-up Fall Assessment

A fall was defined as “the participant unintentionally coming to rest on the floor or some lower level that was not caused by a major intrinsic event.”²²⁾ The incidence of falls was documented when the participant underwent a baseline assessment and during their annual follow-up medical check by asking the participant, “Did you fall during this year?” If the participant did not visit the examination site for a follow-up medical check, we mailed them the questionnaires to determine their incidence of falling. The “fall group” was defined as those who fell at least once during the follow-up period, and the “non-fall group” was defined as those who did not fall during the follow-up period.

Statistical Analysis

A Shapiro–Wilk test was performed to confirm the normal distribution of the data. Normally distributed continuous variables are expressed as mean values, and non-normally distributed continuous variables are expressed as median values. Characteristics were analyzed between the fall group and the non-fall group. Comparisons of continuous variables were made using an unpaired t-test, and those without normality were made using a Mann–Whitney U test. Categorical variables were assessed using a χ² test. A principal component analysis was performed using all screening tests for locomotive dysfunction as variables. The number of principal components was determined for components with eigenvalues greater than 1, and principal component loadings were calculated. The binary regression method was used to assess factors that predict the incidence of a fall during follow-up. The objective variable was the incidence of falls, and the exploratory variables were selected locomotive function screening tests for each principal component. Exploratory variables were entered using a forced entry method with age and past fall history into the model in the binary regression analysis to calculate hazard ratios with 95% confidence intervals. All data were analyzed using SPSS statistical software (Version 26 for Windows; IBM, Tokyo, Japan). Statistical significance was recognized for P < 0.05.

RESULTS

The study flowchart of patient recruitment and participation is shown in Fig. 1. The follow-up survey was completed by 379 (81.2%) participants (148 men, 231 women) ranging in age from 40 to 97 years. During the follow-up period, 82 participants (21.6%) experienced falls, 11 of whom sustained fractures, whereas 297 participants did not experience any falls or fractures. Among those who fell, 29 experienced multiple falls.

Fig. 1.

Study flowchart of patient participation. A total of 379 subjects completed the 1-year follow-up; 84 subjects did not participate in the follow-up survey.

The baseline characteristics of the participants are summarized in Table 1. Age, BMI, use of analgesic medication, presence of eye problems, hearing problems, hypertension, cerebrovascular disease, and previous history of falls were significantly higher in the fall group than in the non-fall group. Mean height was significantly lower in the fall group.

Table 1. Characteristics of fall group and non-fall group at baseline

Characteristic	Total	Fall group	Non-fall group	P value
	(n=379)	(n=82)	(n=297)
Age, years	72.4 (9.1)	75.0 (9.2)	72.0 (8.8)	0.007
Female, n (%)	231 (60.9)	57 (69.5)	174 (58.6)	0.073
Height, cm	155.3 (8.9)	152.0 (8.6)	156.3 (8.8)	<0.001
Weight, kg	54.3 (10.1)	53.4 (9.9)	54.6 (9.9)	0.319
BMI, kg/m2	22.4 (2.9)	22.9 (3.1)	22.2 (2.8)	0.041
Medication,%
Analgesic medication	20.6	29.3	18.2	0.028
Hypnotics	12.1	14.6	11.4	0.434
Diuretics	6.9	9.8	6.1	0.241
Comorbidities, %
Eye problems a	22.2	32.9	19.2	0.008
Hearing problems b	20.3	32.9	16.8	0.001
Hypertension	36.6	48.1	33.3	0.014
Diabetes	7.8	9.9	7.2	0.430
Cardiovascular disease	8.1	6.2	8.6	0.480
Cerebrovascular disease	4.6	9.9	3.1	0.010
Osteoarthritis of the knee	18.5	23.1	17.2	0.215
Osteoarthritis of the hip	2.9	4.9	2.4	0.229
Lumbar disease	15.9	18.3	15.2	0.697
Osteoporosis	12.7	15.9	11.8	0.327
History of falling (past year)	19.3	46.3	11.8	<0.001

Normally distributed continuous variables are expressed as mean (standard deviation). Continuous variables were analyzed using an unpaired t test. Categorical variables were analyzed using a χ² test.

^a Do you have difficulty seeing? Yes/No.

^b Are you hard of hearing? Yes/No.

Table 2 and Fig. 2 show the results of principal component analysis used to divide components using all screening tests for locomotive dysfunction. The primary variable making up the first principal component was “mobility function,” the second was “muscle function,” and the third was related to “spinal alignment.” The cumulative contribution of the three principal components was 64%. The highest principal component loadings for each component were GLFS-5, muscle mass index, and kyphosis index.

Table 2. Results of principal component analysis

Factor	Principal component
	1	2	3
GLFS-5	−0.837a	0.277	0.058
%YAM	0.634a	0.316	−0.117
Gait speed	0.625a	−0.160	0.432
KCL	−0.597a	0.444	0.004
VAS	−0.565a	0.446	0.320
Gait variability	0.408a	−0.243	0.085
Muscle mass index	0.425	0.784b	−0.153
Grip strength	0.624	0.639b	−0.059
Kyphosis index	0.115	0.138	0.896c

^a Contributing factors to mobility function; ^b Contributing factors to muscle function; ^c Contributing factors to spinal alignment.

Fig. 2.

Plot of the components of principal component analysis. The relationships between each screening test for locomotive dysfunction are shown in a three-dimensional grid. The more distant a component is from the central coordinates, the larger the principal component loadings.

Table 3 shows the results of GLFS-5, muscle mass index, and kyphosis index between the fall group and the non-fall group at baseline. The fall group had a significantly higher proportion of participants exceeding the GLFS-5 cutoff value.

Table 3. Baseline proportions of participants above/below thresholds used to diagnose locomotive dysfunction: GLFS-5, muscle mass index, and kyphosis index

Assessment	Total	Fall group	Non-fall group	P value
	(n=379)	(n=82)	(n=297)
GLFS-5 (%)	19.6	34.1	15.5	<0.001
Muscle mass index (%)	21.3	22.0	19.0	0.530
Kyphosis index (%)	34.2	34.1	34.3	0.973

GLFS-5 cutoff >6; muscle mass index cutoff <7.0 kg/m² (male), <5.7 kg/m² (female); kyphosis index cutoff >13%.

Table 4 shows the results of binary regression analyses used to find factors related to the incidence of falling, which included GLFS-5, muscle mass index, and kyphosis index adjusted for age and previous falls. GLFS-5 was independently related with the incidence of falling [odds ratio=2.04; 95% confidence interval (CI): 1.04–3.99].

Table 4. Odds ratios for the incidence of fall using binary regression model

Assessment	Odds ratio	95% CI
GLFS-5	2.04	1.04–3.99
Muscle mass index	1.09	0.53–2.02
Kyphosis index	0.88	0.45–1.71

Adjusted for age and previous fall history. GLFS-5 cutoff >6; muscle mass index cutoff <7.0 kg/m² (male), <5.7 kg/m² (female); kyphosis index cutoff >13%.

DISCUSSION

The aim of this study was to determine the relative importance of locomotive dysfunction assessments during locomotive health checkups and to identify which screening tests are most predictive for the incidence of falls. Principal component analysis of many screening tests for locomotive dysfunction during a scheduled locomotive health checkup revealed that they can be divided into elements related to mobility function, muscle function, and spinal alignment. These three elements appear to be central to the concept of LS.²⁾ The mobility function component can be used to screen for osteoarthritis and motor function diseases, which can be related to LS, whereas the muscle function component is useful for the screening of sarcopenia, and the spinal alignment component is useful for the screening of osteoporosis. We propose that these three components can be used to better confirm the overall locomotive dysfunction when conducting a locomotive health checkup. GLFS-5, muscle mass index, and the kyphosis index had particularly high principal component loadings. Although various assessments of physical and locomotive function are conducted during health checkups to identify older adults who are frail and/or have LS in the community, there are many tests that duplicate the assessed elements of locomotive function. In addition, under circumstances of limited human resources, it is useful to conduct tests that can efficiently identify risk using as few assessments as possible. In this respect, the three tests of apparent highest utility in this study are considerably easy to use.

To our knowledge, this is the first report showing an association between the GLFS-5 score and falls in a prospective study of community-dwelling older adults. However, there have been reports on the risk of postoperative falls in patients with degenerative cervical myelopathy.¹⁶⁾ In a previous prospective study, the GLFS-25 was used to predict falls among older adults in general or patients with disabilities.^23,24) However, completing a 25-item questionnaire can be time-consuming and psychologically burdensome for older adults. We recommend the use of the GLFS-5 because it has the advantage of requiring fewer questions than the GLFS-25, which makes the GLFS-5 easier to administer. Among the questions included in the survey are those asking whether a person can climb stairs and how long they can walk without resting. The GLFS-5 is a method that can easily identify the risk of falling without using exercise tests or tools. In the present study, a cutoff of 6 points for the GLSF-5 was entered as an independent variable in the multivariate analysis and could discriminate between the fall group and the non-fall group. One previous study predicted the incidence of falls during the first postoperative year in patients with degenerative cervical myelopathy, where the cutoff was 12 points for the GLFS-5 score.¹⁶⁾ Therefore, it can be inferred that, unlike in patient groups, the risk of falling in the general elderly population can be identified at a lower cutoff point. However, a study in which the GLFS-5 was developed found that the best cutoff value for identifying LS was a GLFS-5 score of 6 points.⁶⁾ Another study used GLFS-5 cutoff scores to discriminate LS grade 1, LS grade 2, and LS grade 3 patients diagnosed using the GLFS-25 in 1258 community-dwelling participants.²⁵⁾ The study reported that the cutoff scores of the GLFS-5 for identifying the LS grade were 2 for grade 1 (sensitivity 91.7%, specificity 77.8%), 4 for grade 2 (sensitivity 95.7%, specificity 81.7%), and 6 for grade 3 (sensitivity 92.9%, specificity 90.0%). LS grade 3 is a condition in which mobility has declined to the extent that it interferes with social participation, making it as severe as physical frailty. The results of this study suggest that a GLFS-5 score of 6 not only signifies severe LS but also indicates a higher risk of future falls. Using the GLFS-5 as a screening test for locomotive dysfunction in community-dwelling elderly individuals offers a simple and accessible method for assessing fall risk, even for social workers and municipal staff with limited medical knowledge.

The present study has several limitations. Given that the participants were only followed closely for 1 year, the longer-term risk of falls could not be determined. In addition, we could not assess the incidence of falls caused by external or environmental factors. It is also possible that the reported number of falls was lower than the real number because approximately 20% of the participants were lost to follow-up. Furthermore, recall bias and errors in answering questions because of cognitive decline are inevitable when investigating falls. However, this study excluded individuals certified as requiring nursing care and focused on independently living elderly individuals, which we believe minimizes these risks. In future studies, particularly in a cohort study with a larger sample size, it will be necessary to include diseases that showed significant differences between the fall and non-fall groups in the univariate analysis and are known to be associated with falls as independent variables. This would enhance the accuracy and robustness of the model.

CONCLUSION

Principal component analysis of many screening tests for locomotive dysfunction during a locomotive health checkup revealed that tests can be divided into elements related to mobility function, muscle function, and spinal alignment. The tests with high principal component loadings were the GLFS-5, the muscle mass index, and the kyphosis index. Overall, the GLFS-5 is a simple and convenient screening test that we demonstrated can predict falls.

ACKNOWLEDGMENTS

The authors sincerely appreciate the contributions of all staff members involved in the GAINA study. This work was supported by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (Grant Number JP19K11809).

CONFLICTS OF INTEREST

The authors declare no conflict of interest.

REFERENCES

• 1.  Ministry of Health, Labor, and Welfare: Summary Report of Comprehensive Survey of Living Conditions. Government of Japan. 2022. https://www.mhlw.go.jp/toukei/saikin/hw/k-tyosa/k-tyosa22/dl/06.pdf. Accessed 10 Oct 2024.
• 2.  Ishibashi H: Fall risk and fracture. Locomotive syndrome and fall [in Japanese]. Clin Calcium 2013;23:669–677. PMID:23628679
• 3.  Kobayashi K, Ando K, Nakashima H, Machino M, Kanbara S, Ito S, Inoue T, Yamaguchi H, Koshimizu H, Ishiguro N, Hasegawa Y, Imagama S: Overcoming locomotive syndrome: the Yakumo study. Mod Rheumatol 2021;31:750–754. PMID:33492184, DOI:10.1080/14397595.2021.1879413
• 4.  Matsumoto H, Hagino H, Osaki M, Tanishima S, Tanimura C, Matsuura A, Makabe T: Gait variability analysed using an accelerometer is associated with locomotive syndrome among the general elderly population: the GAINA study. J Orthop Sci 2016;21:354–360. PMID:27020175, DOI:10.1016/j.jos.2016.02.003
• 5.  Teshima Y, Miyamoto Y, Kawaguchi T, Yamakawa K, Otsubo Y, Koyanagi N, Yamada Y, Shigemitsu T, Ishida C, Nakashima N: Significance of locomotive medical examination in the ningen dock health checkup [in Japanese]. Ningen Dokku 2021;36:26–31.
• 6.  Seichi A, Hoshino Y, Doi T, Akai M, Tobimatsu Y, Iwaya T: Development of a screening tool for risk of locomotive syndrome in the elderly: the 25-question Geriatric Locomotive Function Scale. J Orthop Sci 2012;17:163–172. PMID:22222445, DOI:10.1007/s00776-011-0193-5
• 7.  Arai H, Satake S: English translation of the Kihon Checklist. Geriatr Gerontol Int 2015;15:518–519. PMID:25828791, DOI:10.1111/ggi.12397
• 8.  Chen LK, Woo J, Assantachai P, Auyeung TW, Chou MY, Iijima K, Jang HC, Kang L, Kim M, Kim S, Kojima T, Kuzuya M, Lee JS, Lee SY, Lee WJ, Lee Y, Liang CK, Lim JY, Lim WS, Peng LN, Sugimoto K, Tanaka T, Won CW, Yamada M, Zhang T, Akishita M, Arai H: Asian Working Group for Sarcopenia: 2019 consensus update on sarcopenia diagnosis and treatment. J Am Med Dir Assoc 2020;21:300–307. PMID:32033882, DOI:10.1016/j.jamda.2019.12.012
• 9.  Moayyeri A, Adams JE, Adler RA, Krieg MA, Hans D, Compston J, Lewiecki EM: Quantitative ultrasound of the heel and fracture risk assessment: an updated meta-analysis. Osteoporos Int 2012;23:143–153. PMID:22037972, DOI:10.1007/s00198-011-1817-5
• 10.  Camozzi V, De Terlizzi F, Zangari M, Luisetto G: Quantitative bone ultrasound at phalanges and calcaneus in osteoporotic postmenopausal women: influence of age and measurement site. Ultrasound Med Biol 2007;33:1039–1045. PMID:17445968, DOI:10.1016/j.ultrasmedbio.2007.01.002
• 11.  Bet P, Castro PC, Ponti MA: Fall detection and fall risk assessment in older person using wearable sensors: a systematic review. Int J Med Inform 2019;130:103946. PMID:31450081, DOI:10.1016/j.ijmedinf.2019.08.006
• 12.  McDaniels-Davidson C, Davis A, Wing D, Macera C, Lindsay SP, Schousboe JT, Nichols JF, Kado DM: Kyphosis and incident falls among community-dwelling older adults. Osteoporos Int 2018;29:163–169. PMID:29018904, DOI:10.1007/s00198-017-4253-3
• 13.  Asahi R, Nakamura Y, Koike Y, Kanai M, Watanabe K, Yuguchi S, Kamo T, Azami M, Ogihara H, Asano S: Does locomotive syndrome severity predict future fragility fractures in community-dwelling women with osteoporosis? Mod Rheumatol 2023;33:1036–1043. PMID:36029035, DOI:10.1093/mr/roac101
• 14.  Matsumoto H, Tanimura C, Tanishima S, Osaki M, Noma H, Hagino H: Sarcopenia is a risk factor for falling in independently living Japanese older adults: a 2‐year prospective cohort study of the GAINA study. Geriatr Gerontol Int 2017;17:2124–2130. PMID:28517243, DOI:10.1111/ggi.13047
• 15.  Tominaga R, Fukuma S, Yamazaki S, Sekiguchi M, Otani K, Kikuchi S, Sasaki S, Kobayashi S, Fukuhara S, Konno S: Relationship between kyphotic posture and falls in community-dwelling men and women: the locomotive syndrome and health outcome in Aizu cohort study. Spine 2016;41:1232–1238. PMID:27046636, DOI:10.1097/BRS.0000000000001602
• 16.  Kimura A, Takeshita K, Inoue H, Inose H, Yoshii T, Maekawa A, Endo K, Miyamoto T, Furuya T, Nakamura A, Mori K, Seki S, Kanbara S, Imagama S, Matsunaga S, Yamazaki M, Okawa A: The 5-question Geriatric Locomotive Function Scale predicts postoperative fall risk in patients undergoing surgery for degenerative cervical myelopathy. J Orthop Sci 2021;26:779–785. PMID:33109435, DOI:10.1016/j.jos.2020.08.017
• 17.  Matsumoto H, Tanimura C, Tanishima S, Hagino H: Association between speed of sound of calcaneal bone assessed by quantitative ultrasound and sarcopenia in a general older adult population: a cross-sectional study. J Orthop Sci 2019;24:906–911. PMID:30733130, DOI:10.1016/j.jos.2019.01.003
• 18.  Noguchi T, Murata C, Hayashi T, Watanabe R, Saito M, Kojima M, Kondo K, Saito T: Association between community-level social capital and frailty onset among older adults: a multilevel longitudinal study from the Japan Gerontological Evaluation Study (JAGES). J Epidemiol Community Health 2022;76:182–189. PMID:34341149, DOI:10.1136/jech-2021-217211
• 19.  Matsumoto H, Makabe T, Morita T, Ikuhara K, Kajigase A, Okamoto Y, Ashikawa E, Kobayashi E, Hagino H: Accelerometry-based gait analysis predicts falls among patients with a recent fracture who are ambulatory: a 1-year prospective study. Int J Rehabil Res 2015;38:131–136. PMID:25486217, DOI:10.1097/MRR.0000000000000099
• 20.  Morita T, Matsumoto H, Makabe T, Hagino H: Influence of standing posture on physical function and incidence of multiple falls in community-dwelling older adults [in Japanese]. Rigakuryoho Kagaku 2021;36:499–504.
• 21.  Teragaki Y, Araya K, Sueki K, Iriuchijima K, Yamauchi J, Usuda S: Reliability and validity of the index of kyphosis on sitting [in Japanese]. Rigakuryoho Kagaku 2004;19:137–140.
• 22.  Kellogg International Work Group . Dan Med Bull 1987;34:1–24. PMID:3595217
• 23.  Iida H, Seki T, Takegami Y, Osawa Y, Kato D, Takemoto G, Ando K, Ishizuka S, Hasegawa Y, Imagama S: Association between locomotive syndrome and fall risk in the elderly individuals in Japan: the Yakumo study. J Orthop Sci 2024;29:327–333. PMID:36522244, DOI:10.1016/j.jos.2022.11.023
• 24.  Kimura A, Takeshita K, Inoue H, Seichi A, Kawasaki Y, Yoshii T, Inose H, Furuya T, Takeuchi K, Matsunaga S, Seki S, Tsushima M, Imagama S, Koda M, Yamazaki M, Mori K, Nishimura H, Endo K, Yamada K, Sato K, Okawa A: The 25-question Geriatric Locomotive Function Scale predicts the risk of recurrent falls in postoperative patients with cervical myelopathy. J Orthop Sci 2018;23:185–189. PMID:29100824, DOI:10.1016/j.jos.2017.10.006
• 25.  Kobayashi T, Morimoto T, Shimanoe C, Ono R, Otani K, Mawatari M: Development of a simple screening tool based on the 5-question Geriatric Locomotive Function Scale for locomotive syndrome. J Orthop Sci 2022;27:913–920. PMID:34090778, DOI:10.1016/j.jos.2021.05.001