Original Article

Yonsei Med J 2018 Mar;59(2):317-324

https://doi.org/10.3349/ymj.2018.59.2.317 pISSN: 0513-5796 · eISSN: 1976-2437

Prevalence of Frailty in Patients with Osteoporotic

Vertebral Compression Fracture and Its Association
with Numbers of Fractures
Ho-Joong Kim1*, Saejong Park2*, Soo-Hyun Park2, Jiwon Park1,
Bong-Soon Chang3, Choon-Ki Lee3, and Jin S. Yeom1
1
Spine Center and Department of Orthopedic Surgery, Seoul National University College of Medicine, Seoul National University
Bundang Hospital, Seongnam;
2
Division of Sports Science, Korea Institute of Sports Science, Seoul;
3
Department of Orthopaedic Surgery, Seoul National University College of Medicine, Seoul National University Hospital, Seoul, Korea.

Purpose: To assess the association between frailty and osteoporotic vertebral compression fracture (OVCF) and to evaluate the
relationship between numbers of OVCFs and frailty.
Materials and Methods: We enrolled 760 subjects, including 59 patients (with OVCF) and 701 controls (without OVCF). Success-
ful matching provided 56 patient-control pairs. We analyzed principal clinical and demographic information, which included sex,
age, height, weight, body mass index (BMI), variable frailty phenotypes, and Oswestry Disability Index (ODI) and EuroQol 5-di-
mension questionnaire (EQ-5D) scores. The association between frailty and OVCF was ascertained. In addition, the degrees of
disability and quality of life attributable to frailty were determined.
Results: The prevalence of frailty was significantly higher in the OVCF group than in the control group (p<0.001). Most of the frail-
ty phenotypes, such as exhaustion, physical inactivity, slowness, and handgrip strength, were also significantly observed in the
OVCF group. Within the OVCF group, the participants with frailty had significantly higher disability and lower quality of life than
those in a robust state (p<0.001 for ODI and EQ-5D). In addition, the multivariate logistic regression analysis demonstrated that
the patients with low BMI [odds ratio (OR)=0.704; 95% confidence interval (CI), 0.543−0.913] and ≥3 fractures (OR=9.213; 95% CI,
1.529−55.501) within the OVCF group were associated with higher odds of frailty.
Conclusion: The present study showed significant relationships between frailty and OVCF, severity of symptoms, and disability
induced by OVCF. Furthermore, frailty could be a causal and/or resulting factor of OVCFs.

Key Words: Frailty, osteoporotic vertebral compression fracture, Oswestry disability index, EuroQol 5-dimension questionnaire,
Fried frailty criteria

INTRODUCTION
Received: November 2, 2017 Revised: December 30, 2017
Accepted: January 8, 2018 As the mean age of the general population is increasing, one
Corresponding author: Dr. Jin S. Yeom, Spine Center and Department of Orthopedic
Surgery, Seoul National University College of Medicine, Seoul National University
of the most problematic health outcomes is the clinical condi-
Bundang Hospital, 82 Gumi-ro 173beon-gil, Bundang-gu, Seongnam 13620, Korea. tion of frailty.1,2 Growing research has deemed frailty a serious
Tel: 82-31-787-7195, Fax: 82-31-787-4056, E−mail: [email protected] and increasing public health issue, which develops as a con-
*Ho-Joong Kim and Saejong Park contributed equally to this work. sequence of aging-related decline in many physiological sys-
•The authors have no financial conflicts of interest. tems, including physical, psychological, and social functions,
© Copyright: Yonsei University College of Medicine 2018 collectively resulting in a clinical condition with an increased
This is an Open Access article distributed under the terms of the Creative Com- vulnerability to sudden changes in health status triggered by
mons Attribution Non−Commercial License (http://creativecommons.org/licenses/
by−nc/4.0) which permits unrestricted non−commercial use, distribution, and re- minor stressor events.2,3 This increases the risk of adverse out-
production in any medium, provided the original work is properly cited. comes, including falls, delirium, disability, and mortality.1-3

www.eymj.org 317
 Frailty and Osteoporotic Vertebral Fracture

Previous studies have shown that osteoporotic fractures are a minor trauma at least 6 months prior. We defined osteopo-
associated with frailty, and frailty is further worsened after os- rotic vertebral fracture as an axial compression of the verte-
teoporotic fractures because of deficit accumulation being bral body with intact posterior constraining elements, which
greater.4,5 In addition, a systematic review reported that frailty included wedge, biconcavity, and compression deformity as
and pre-frailty are significant predictors of osteoporotic frac- described by Eastell, el al.8 The inclusion criteria for the healthy
tures in older adults.6 Given that osteoporotic vertebral com- control group were as follows: age of 65 to 85 years, absence of
pression fracture (OVCF) is the most common osteoporotic fra- low back pain, and no history of OVCF. The main exclusion
gility fracture in older adults,5 a possible association may exists criteria for both groups were as follows: any neurological defi-
between OVCF and frailty. Nonetheless, frailty in patients with cit caused by OVCF; severe joint pain impeding walking; pres-
vertebral fragility fractures has not been robustly studied. ence of peripheral vascular diseases; any clinically significant
We hypothesized that OVCF would be significantly associ- medical comorbidity, such as sepsis, which might influence
ated with frailty and that a positive correlation exists between the general medical condition of the patients; and cancer. Study
frailty and numbers of OVCFs. The purpose of this study was to subjects with incomplete questionnaire findings were exclud-
assess the association between frailty and OVCF and to evalu- ed from the study.
ate the relationship between numbers of OVCFs and frailty.
Data collection
For each subject, the following baseline clinical and demo-
MATERIALS AND METHODS graphic variables were collected: sex, age, height, weight, body
mass index (BMI), variables regarding frailty assessment,
Study design and study population medical history, and clinical outcomes, including the Oswes-
This propensity score-matched case-control study was per- try Disability Index (ODI) and the EuroQol 5-dimension ques-
formed within the framework of a prospective study designed tionnaire (EQ-5D).9-11 Bone mineral densities at the lumbar
to develop criterion-referenced health-related fitness standards spine and hip joints were measured using a dual-energy ab-
for the National Fitness Award.7 This study was approved by sorptiometry scan. In addition, radiographic images were ob-
the Institutional Review Boards of Seoul National University tained in the standing position.
Bundang Hospital and Korea Institute of Sport Science, and
was conducted in accordance with the approved study proto- Frailty measurement
col (IRB No. B-1612-373-301). All participants provided writ- Frailty was defined in accordance with the following five phe-
ten informed consent before study participation. For the study, notypes as reported by Fried, et al.12 (Table 1): weight loss, ex-
760 participants were enrolled, including 59 patients with a haustion, physical inactivity, slowness, and handgrip strength.
diagnosis of OVCF (OVCF group) and 701 healthy individuals
without any osteoporotic compression fracture or back pain Weight loss
(control group). The patients with OVCF were recruited from a Weight loss was defined as a loss of >5 kg in the prior year.
single center of a tertiary-care teaching hospital and the con-
trol subjects from either the National Fitness Award or com- Exhaustion
munity centers from August 2014 to February 2017. The eligi- Exhaustion was assessed using the following question from
bility criteria for the OVCF group were as follows: age of 65 to the Center for Epidemiological Studies-Depression Scale:12 A)
85 years and an old vertebral compression fracture caused by ‘‘I felt that everything I did was an effort’’ and B) ‘‘I could not

Table 1. Frailty-Defining Criteria Used in this Study

Frailty phenotype Definition and cutoff criteria
Weight loss Weight loss was defined as a loss of more than 5 kg in the last year
Self-report of either of the following: (i) felt that everything I did was an effort in the last week or (ii) I could not get going
Exhaustion in the last week; 0=rarely or none of the time (1 day), 1=some or a little of the time (1–2 days), 2=a moderate amount
of the time (3–4 days), and 3=most of the time. Subjects who answered “2” or “3” were categorized as frail.
Physical inactivity The lowest sex-specific 20% of the study population (IPAQ score)
Slowness 5-meter walk at one’s usual pace: lowest sex-specific 20% of the study population
BMI (kg/m2)-male Cutoff value (kg) BMI (kg/m2)-female Cutoff value (kg)
≤24 ≤29 ≤23 ≤17
Handgrip strength 24–28 ≤30 23–26 ≤17.3
26–29 ≤18
>28 ≤32 >29 ≤21
BMI, body mass index; IPAQ, International Physical Activity Questionnaire.

318 https://doi.org/10.3349/ymj.2018.59.2.317
Ho-Joong Kim, et al.

get going.’’ If answered yes, the following question was asked: this study, one-to-one matching was used with a nearest neigh-
‘‘How often in the previous week did you feel this way?’’ Scores bor matching without replacements, in which each case in the
ranged from 0 to 3, where 0 indicated rarely or none of the time control group was matched with a unique case in the OVCF
(1 day), 1 indicated some of the time (1−2 days), 2 indicated a group based on the nearest propensity scores. Factors associ-
moderate amount of the time (3−4 days), and 3 indicated most ated with frailty, such as age, sex, BMI, education level, and in-
of the time. The subjects answering 2 or 3 to either of these come, were considered as confounders, and used in the logis-
questions were categorized as frail, as per exhaustion criterion. tic regression analysis.2,15-17 This procedure produced 56 well-
matched pairs in the OVCF and control groups. Three subjects
Physical inactivity in the OVCF could not be matched with those in the control
For physical activity, the International Physical Activity Ques- group because of a fairly large discrepancy in the propensity
tionnaire-Short Form was used.13 The participants were asked scores of three subjects between both groups.
regarding the amount of time they spent engaged in physical In the propensity score-adjusted groups, the baseline clinical
activities during the past week, and those in the lowest quintile and demographic variables were analyzed using the t test and
were defined as physically inactive. chi-square test for categorical and continuous data analyses.
The adjusted prevalence of frailty was compared between the
Slowness OVCF and control groups. Given the difference in the preva-
Slowness was defined as the slowest quintile of the walking per- lence of frailty between the two groups, the post hoc power
formance of the subjects at 5 m. Adjustments were made for the was also calculated. The Kruskal-Wallis test with post hoc
sex and standing height of the patients. Bonferroni-corrected Mann-Whitney tests was used for the
comparison of the clinical outcomes and other variables with-
Handgrip strength in each study group. Furthermore, the analysis of covariance
For the handgrip strength measurement, the participants were was conducted to stratify age and sex.
instructed to squeeze a handgrip dynamometer (GRIP-D5101; In the OVCF group, the chi-square test was used to analyze
Takei, Niigata, Japan) as hard as possible; this exercise was re- the association between the numbers of OVCF and frailty. To
peated thrice (once with each hand and then with the strongest adjust for the confounders, such as age, BMI, and sex, univari-
hand), and the maximum value was recorded. Sex- and BMI- ate and multivariate models were created. The variables that
specific cutoff values for grip strength were used to identify were significantly associated with frailty at p values of <0.10 in
the subjects with frailty.12 The subjects who did not fulfill any the univariate analysis were entered in the multivariate model,
criteria for frailty were considered robust, and those who ful- along with potentially important variables, including age, BMI,
filled one or two criteria were considered pre-frail. If three or and sex, regardless of their statistical significance. For the
more frail phenotypes existed in the subjects, then they were multivariate model, we anticipated a potential issue of collin-
considered frail. earity between the variables and set an a priori rule to exclude
variables with correlation coefficients of ≥0.50. The alpha level
Clinical outcome variables of significance was set at 0.05. All statistical analyses were per-
The ODI is a self-administered questionnaire that measures formed using the SPSS version 20.0 software (IBM Corp., Ar-
back-specific function on a 10-item scale with six response cat- monk, NY, USA), except the post hoc power analysis. Given the
egories for each item. Each item is scored from 0 to 5, and the difference in the prevalence of frailty between the groups, a
summation of scores for each item is converted into a 0−100 post hoc power analysis was also performed in 56 matched
scale.9 The EQ-5D-5L is a five-dimensional health-state classi- cases and controls, with an alpha value of 0.05 using G*power
fication.11 The five dimensions are mobility, self-care, usual ac- 3.1 (Faul, Erdfelder, Lang, & Buchner, 2007).
tivities, pain/discomfort, and anxiety/depression. An EQ-5D-
5L health status is defined by selecting one level from each
dimension. The EQ-5D-5L preference-based measure can be RESULTS
regarded as a continuous outcome scored in a scale of 0−1.00,
with 1.00 indicating full health and 0 representing death.11 Unadjusted case-control group
In total, 760 subjects were recruited in the study, including 59
Statistical analysis patients in the OVCF group and 701 individuals in the control
The t-test and chi-square test were used to analyze unadjusted group (Fig. 1). Table 2 demonstrates the baseline characteris-
continuous and categorical variables. To adjust the differences tics of the participants in both groups. Significant differences
in the baseline characteristics and influential factors for frailty, were found in sex distribution and income (p<0.001 and p=
a propensity score-matched algorithm was used. The propen- 0.022, respectively). As expected, the clinical outcomes, in-
sity score was estimated using a logistic regression model to cluding the ODI and EQ-5D findings, were significantly worse
balance the baseline covariates between the two groups.14 For in the OVCF group than in the control group (p<0.001 for both

https://doi.org/10.3349/ymj.2018.59.2.317 319
 Frailty and Osteoporotic Vertebral Fracture

76 and 701 participants were assessed for eligibility in the OVCF group and control group, respectively

17 patients were excluded in the OVCF group

11 patients declined to participate
6 patients had pain in other joints

59 and 701 participants were enrolled in the OVCF group and control group, respectively

Propensity score matching

OVCF group (n=56) Control group (n=56)

Fig. 1. Enrollment, group assignment, and propensity score matching. OVCF, osteoporotic vertebral compression fracture.

Table 2. Descriptive Statistics of the Subjects in the Unadjusted Population

Demographic data OVCF group (59) Control group (701) p value
Age (yr) 73.5±6.2 71.0±4.3 0.004
Male:Female 7:52 354:347 <0.001
BMI (kg/m2) 24.7±3.0 24.3±2.9 0.390
ODI 33.7±15.6 9.6±11.9 <0.001
EQ-5D 0.564±0.238 0.834±0.151 <0.001
Level of education, n (%) 0.769
Nothing 5 (8.5) 42 (6.0)
Elementary school 20 (33.9) 221 (31.5)
Middle school 9 (15.3) 140 (20.0)
High school 15 (25.4) 194 (27.7)
University 10 (16.9) 104 (14.8)
Education (yr) 10.0±4.2 9.1±4.5 0.478
Income, n (%) 0.022
Low 55 (93.2) 480 (68.5)
Middle 4 (6.8) 184 (26.2)
High 0 37 (5.3)
Living condition, n (%) 0.816
Living alone 12 (20.3) 134 (19.1)
Living with someone 47 (79.7) 567 (80.9)
Smoking, n (%) 0.077
None or past smoker 58 (98.3) 628 (89.6)
Current smoker or ex-smoker within 1 year 1 (1.7) 73 (10.4)
Chronic disease, n (%)
Hypertension 14 (23.7) 297 (42.4)
Diabetes 8 (14.3) 118 (16.8)
Stroke 0 (0) 30 (4.3)
Ischemic heart disease 0 (0) 21 (3.0)
Lung disease 0 (0) 34 (4.9)
Musculoskeletal 8 (14.3) 136 (19.4)
Frailty, n (%) <0.001
Robust 10 (16.9) 372 (53.1)
Pre-frail 25 (42.4) 303 (43.2)
Frail 24 (40.7) 26 (3.7)
OVCF, osteoporotic vertebral compression fracture; BMI, body mass index; SD, standard deviation; ODI, Oswestry Disability Index; EQ-5D, EuroQol 5-dimension
questionnaire.
Frailty was defined using the Fried frailty criteria. Values are presented as mean±SD or number (%).

320 https://doi.org/10.3349/ymj.2018.59.2.317
Ho-Joong Kim, et al.

variables). Furthermore, the subjects in the OVCF group had Propensity score-matched case-control group
a significantly higher percent ratio of frail and pre-frail states The propensity score matching yielded 56 well-matched pa-
than the control group (p<0.001). In the OVCF group, all pa- tients with OVCF and control pairs (Fig. 1). After matching, no
tients had a vertebral fracture from T7 to L5, and the most fre- significant difference was found between the two groups in
quent fractured level was L1 in 27 patients. Twenty-eight pa- terms of age, sex, BMI, educational level, and income. How-
tients (50%) had multiple OVCFs. ever, the OVCF group demonstrated significantly higher dis-
ability (ODI) and lower health-related quality of life (EQ-5D
Table 3. Descriptive Statistics of the Subjects in the Propensity Score-Matched Case-Control Group
Propensity score-matched Propensity score-matched
Demographic data p value
OVCF group (56) control group (56)
Age (yr) 73.1±6.2 72.3±5.0 0.472
Male:Female 7:49 8:48 0.782
BMI (kg/m2) 24.8±3.0 25.0±3.0 0.678
ODI 33.6±15.8 12.7±12.9 <0.001
EQ-5D 0.564±0.240 0.813±0.142 <0.001
Level of education, n (%) 0.642
Nothing 5 (8.9) 6 (10.7)
Elementary school 20 (35.7) 13 (23.2)
Middle school 9 (16.1) 9 (16.1)
High school 15 (26.8) 20 (35.7)
University 7 (12.5) 8 (14.3)
Education (yr) 9.6±4.0 9.0±4.8 0.472
Income, n (%) 0.922
Low 52 (92.9) 53 (92.3)
Middle 4 (7.1) 3 (7.7)
High 0 0
Living condition, n (%) 0.656
Living alone 11 (19.6) 13 (23.2)
Living with someone 45 (80.4) 43 (76.8)
Smoking, n (%) 1.000
Nonsmoker or past smoker 55 (98.2) 54 (96.4)
Current smoker or ex-smoker within 1 year 1 (1.8) 2 (3.6)
Chronic disease, n (%)
Hypertension 15 (26.8) 27 (48.2)
Diabetes 8 (14.3) 8 (14.3)
Stroke 0 (0) 3 (5.4)
Ischemic heart disease 0 (0) 3 (5.4)
Lung disease 0 (0) 4 (7.1)
Musculoskeletal 8 (14.3) 13 (23.2)
Frailty, n (%) <0.001
Robust 10 (17.9) 30 (53.6)
Pre-frail 22 (39.3) 24 (42.9)
Frail 24 (42.9) 2 (3.6)
Frailty criteria, n (%)
Weight loss 17 (30.4) 13 (23.2) 0.393
Exhaustion 26 (46.4) 3 (5.4) <0.001
Physical inactivity 14 (32.6) 7 (12.5) 0.016
Slowness 30 (55.6) 5 (9.4) <0.001
Handgrip strength 53 (37.3) 22 (15.5) <0.001
OVCF, osteoporotic vertebral compression fracture; BMI, body mass index; SD, standard deviation; ODI, Oswestry Disability Index; EQ-5D, EuroQol 5-dimension
questionnaire.
Frailty was defined using the Fried frailty criteria. Values are presented as mean±SD or number (%).

https://doi.org/10.3349/ymj.2018.59.2.317 321
 Frailty and Osteoporotic Vertebral Fracture

findings; p<0.001 for both variables) than the control group. Table 4. Clinical Outcome Variables according to Frailty Status in Each
Table 3 shows that the OVCF group had a significantly high- Group
er prevalence of frailty than the control group (p<0.001); 24 OVCF group (56) Control group (56)
(42.9%) and 22 participants (39.3%) in the OVCF group and 2 p<0.001 p=0.056
ODI
(3.6%) and 24 participants (42.9%) in the control group were (<0.001)* (0.145)*
considered frail and pre-frail, respectively (Table 3). The post Robust 17.5±7.7 10.3±11.3
hoc power analysis confirmed this difference in mean and Pre-frail 30.5±13.8 13.4±12.8
standard deviation in the ratio of frail to pre-frail and robust Frail 42.3±14.1 40.0±5.7
participants, with an alpha value of 0.05 and a statistical power p=0.016 p=0.343
EQ-5D
of 100.0%. Among the frailty criteria, no significant difference (0.027)* (0.802)*
in weight loss was found between the two groups, whereas the Robust 0.713±0.080 0.819±0.132
other criteria were significantly different between them (Table 3). Pre-frail 0.616±0.268 0.818±0.153
Frail 0.463±0.214 0.658±0.116
Within-group analysis for the relationship between OVCF, osteoporotic vertebral compression fracture; ODI, Oswestry Disability
frailty and clinical outcomes Index; EQ-5D, EuroQol 5-dimension questionnaire.
Within the OVCF group, the participants with frailty had sig- *p values adjusted for age and sex using analysis of covariance.
nificantly higher disability and lower quality of life than those
Table 5. Association between the Numbers of OVCFs and Frailty and
in a robust state (p<0.001 for ODI and EQ-5D) (Table 4). The Clinical Outcomes
post hoc analysis with Bonferroni correction for the ODI and
Numbers Frailty, n (%)
EQ-5D scores showed significant differences between the frail p value
of OVCF Frail Pre-frail Robust
and pre-frail patients (p=0.034 and p=0.032, respectively) and
1 9 (32.1) 15 (53.6) 4 (14.3)
between the frail and robust patients (p<0.001 and p=0.004, re-
2 6 (40.0) 3 (20.0) 6 (40.0) 0.013
spectively) in the OVCF group. After adjustments for age and
≥3 9 (69.2) 4 (30.8) 0 (0)
sex, the ODI and EQ-5D were significantly different with re-
OVCF, osteoporotic vertebral compression fracture.
gard to the frailty status in the OVCF group (Table 4).
been underestimated, compared with that in patients with os-
Numbers of OVCFs and frailty teoporotic hip fractures. The Global Longitudinal Study of Os-
Table 5 demonstrates the association between the numbers of teoporosis in Women (GLOW) study demonstrated that frailty,
OVCFs and frailty. The patients who had ≥3 vertebral fractures according to the Fried frailty criteria, appeared to be associated
had a significantly higher risk for frailty (p=0.013) in the OVCF with fragility fractures, disabilities, and falls.5,18 However, this
group. In the univariate logistic model, consisting of the OVCF study included only 2% of their total cohort patients with exist-
group, education level, income, living status, alcohol habitus, ing vertebral fractures.18 Two previous studies investigated the
smoking, and any kind of chronic disease did not have signifi- association between vertebral compression fracture and frail-
cant association with frailty at a p value of <0.1. Therefore, the ty.5,19 Kado, et al.19 reported that older women with incident
final multivariate logistic analysis was performed in the OVCF vertebral fractures have an increased risk of mortality, which
group so that the impact of age, sex, BMI, and number of frac- can be explained by weight loss and physical frailty. However,
tures could be assessed. It demonstrated that BMI [p=0.008; their study did not use any specific frailty diagnostic criteria,
odds ratio (OR), 0.704; 95% confidence interval (CI), 0.543−0.913] but included only physical frailty markers, such as weight loss,
and the numbers of OVCFs (≥3 fractures vs. <3 fractures; p= inability to rise from a chair, and difficulty standing on the feet
0.015; OR, 9.213; 95% CI, 1.529−55.501) were significantly as- for 2 hours.19 A recent study by Walters, et al.5 showed that frail-
sociated with lower and higher odds of frailty, respectively. ty is prevalent in patients hospitalized owing to vertebral fra-
gility fractures.
The within-group analysis revealed that frailty had a stron-
DISCUSSION ger relationship with more severe symptoms and higher dis-
ability induced by OVCF. This result can also be explained by
The study shows that frailty is prevalent in the patients with the fact that frailty might directly aggravate the disability and
OVCF. More than 50% of the patients in the OVCF group were health-related quality of life of patients with OVCF. The pres-
classified as “frail” according to the Fried frailty criteria. In the ent study could not show a causal relationship between OVCF
OVCF group, the patients with frailty had higher disability and and frailty owing to its cross-sectional design. Nonetheless, we
lower health-related quality of life. Furthermore, the numbers can surmise the relationship from previous studies.4,18,19 In fact,
of OVCF were significantly related to frailty. frailty is well known to be a significant risk factor of fall and
Although frailty has emerged as a significant issue in older osteoporotic fracture.6,18,20,21 Kado, et al.19 reported that physi-
adults, frailty in patients with vertebral fragility fractures has cal frailty may be a risk factor of vertebral fractures, but is like-

322 https://doi.org/10.3349/ymj.2018.59.2.317
Ho-Joong Kim, et al.

ly a complication that occurs after vertebral fractures. A study ORCID

using data from the GLOW Hamilton 3-year cohort clearly
showed that the increase in frailty index was significantly Ho-Joong Kim https://orcid.org/0000-0002-8205-4648
greater in the participants with major osteoporotic fractures Saejong Park https://orcid.org/0000-0001-7229-5790
Jin S. Yeom https://orcid.org/0000-0002-8241-5335
than in those without major osteoporotic fractures, indicating
increasing frailty incidences and faster deficit accumulation
in older women after major osteoporotic fractures.4 Therefore, REFERENCES
based on the results of the present and previous studies, we
1. Keevil VL, Romero-Ortuno R. Ageing well: a review of sarcopenia
consider that frailty causes or results from incident vertebral and frailty. Proc Nutr Soc 2015;74:337-47.
fractures.4,18-20 2. Clegg A, Young J, Iliffe S, Rikkert MO, Rockwood K. Frailty in el-
In addition, the increased numbers of OVCFs were signifi- derly people. Lancet 2013;381:752-62.
cantly associated with a higher prevalence of frailty in the 3. Cooper Z, Rogers SO Jr, Ngo L, Guess J, Schmitt E, Jones RN, et al.
Comparison of frailty measures as predictors of outcomes after
OVCF group. After adjustments for age, sex, and BMI, ≥3 OVCFs
orthopedic surgery. J Am Geriatr Soc 2016;64:2464-71.
were associated with higher odds for frailty than a single num- 4. Li G, Papaioannou A, Thabane L, Cheng J, Adachi JD. Frailty
ber of fracture. This result is in line with the results of a previ- change and major osteoporotic fracture in the elderly: data from
ous study that used the GLOW database,4 in which the increase the global longitudinal study of osteoporosis in women 3-year
in the frailty index was significantly greater in the older wom- hamilton cohort. J Bone Miner Res 2016;31:718-24.
en experiencing a major osteoporotic fracture than in their peer 5. Walters S, Chan S, Goh L, Ong T, Sahota O. The prevalence of frailty
in patients admitted to hospital with vertebral fragility fractures.
controls. Therefore, we consider that OVCF has an accumula- Curr Rheumatol Rev 2016;12:244-7.
tive effect on increasing frailty incidences and worsening defi- 6. Kojima G. Frailty as a predictor of fractures among community-
cits. However, patients with frailty might have higher odds for dwelling older people: a systematic review and meta-analysis.
multiple compression fractures. Bone 2016;90:116-22.
This study had several limitations. First, the sample size in 7. Park S, Kim HJ, Ko BG, Chung JW, Kim SH, Park SH, et al. The
prevalence and impact of sarcopenia on degenerative lumbar spi-
the OVCF group was relatively smaller than that in the control
nal stenosis. Bone Joint J 2016;98-B:1093-8.
group. Although the post hoc power was 100%, such a relative- 8. Eastell R, Cedel SL, Wahner HW, Riggs BL, Melton LJ 3rd. Classi-
ly small sample size in the OVCF group might have led to a se- fication of vertebral fractures. J Bone Miner Res 1991;6:207-15.
lection bias. Second, the present study could not reveal the 9. Fairbank JC, Pynsent PB. The Oswestry Disability Index. Spine
causal relationship between OVCF and frailty, owing to the (Phila Pa 1976) 2000;25:2940-52.
10. Dolan P. Modeling valuations for EuroQol health states. Med Care
cross-sectional design of the study. However, we think that
1997;35:1095-108.
OVCF would be a cause and/or result of frailty. Future longi- 11. Rabin R, de Charro F. EQ-5D: a measure of health status from the
tudinal studies may provide some values to the existing evi- EuroQol Group. Ann Med 2001;33:337-43.
dence on frailty and OVCF. Third, the relationship between 12. Fried LP, Tangen CM, Walston J, Newman AB, Hirsch C, Gottdiener
sagittal alignment, including lumbar lordosis and kyphosis, and J, et al.; Cardiovascular Health Study Collaborative Research
Group. Frailty in older adults: evidence for a phenotype. J Geron-
ODI/EQ-5D should have been analyzed. However, because
tol A Biol Sci Med Sci 2001;56:M146-56.
radiography was obtained only at the fracture level, the global 13. Chun MY. Validity and reliability of Korean version of interna-
and focal alignment could not be measured. tional physical activity questionnaire short form in the elderly.
In conclusion, the present study demonstrates a significant Korean J Fam Med 2012;33:144-51.
association between OVCF and frailty. Clinical outcomes, 14. Kim RB, Garcia RM, Smith ZA, Dahdaleh NS. Impact of resident
such as disability and health-related quality of life, were sig- participation on outcomes after single-level anterior cervical dis-
kectomy and fusion: an analysis of 3265 patients from the ameri-
nificantly associated with frailty in the OVCF group. Therefore, can college of surgeons national surgical quality improvement
as hip fractures have traditionally been regarded to represent program database. Spine (Phila Pa 1976) 2016;41:E289-96.
frailty,4,6,18,21,22 proper attention and management of both frail- 15. Theou O, Brothers TD, Rockwood MR, Haardt D, Mitnitski A,
ty and OVCF are necessary because both have reciprocal in- Rockwood K. Exploring the relationship between national eco-
teraction, that is worsening frailty deficits by fracture and ac- nomic indicators and relative fitness and frailty in middle-aged
and older Europeans. Age Ageing 2013;42:614-9.
celerated risk of OVCF by frailty.
16. Chang YW, Chen WL, Lin FG, Fang WH, Yen MY, Hsieh CC, et al.
Frailty and its impact on health-related quality of life: a cross-sec-
ACKNOWLEDGEMENTS tional study on elder community-dwelling preventive health ser-
vice users. PLoS One 2012;7:e38079.
This study was supported by research grants from the Korea 17. Hoogendijk EO, van Hout HP, Heymans MW, van der Horst HE,
Frijters DH, Broese van Groenou MI, et al. Explaining the associa-
Sports Promotion Foundation, Basic Science Research Program
tion between educational level and frailty in older adults: results
through the National Research Foundation of Korea (NRF) from a 13-year longitudinal study in the Netherlands. Ann Epide-
funded by the Ministry of Education, Science and Technology miol 2014;24:538-44.
(2016R1A2B3012850), and research grants from Hanlim Pharm. 18. Tom SE, Adachi JD, Anderson FA Jr, Boonen S, Chapurlat RD,

https://doi.org/10.3349/ymj.2018.59.2.317 323
 Frailty and Osteoporotic Vertebral Fracture

Compston JE, et al.; GLOW Investigators. Frailty and fracture, dis- Study. J Am Geriatr Soc 2005;53:1321-30.
ability, and falls: a multiple country study from the global longitu- 21. Liu LK, Lee WJ, Chen LY, Hwang AC, Lin MH, Peng LN, et al. As-
dinal study of osteoporosis in women. J Am Geriatr Soc 2013;61: sociation between frailty, osteoporosis, falls and hip fractures
327-34. among community-dwelling people aged 50 years and older in
19. Kado DM, Duong T, Stone KL, Ensrud KE, Nevitt MC, Greendale taiwan: results from I-Lan Longitudinal Aging Study. PLoS One
GA, et al. Incident vertebral fractures and mortality in older wom- 2015;10:e0136968.
en: a prospective study. Osteoporos Int 2003;14:589-94. 22. Tucker A, Donnelly KJ, McDonald S, Craig J, Foster AP, Acton JD.
20. Woods NF, LaCroix AZ, Gray SL, Aragaki A, Cochrane BB, Brun- The changing face of fractures of the hip in Northern Ireland: a
ner RL, et al. Frailty: emergence and consequences in women 15-year review. Bone Joint J 2017;99-B:1223-31.
aged 65 and older in the Women’s Health Initiative Observational

324 https://doi.org/10.3349/ymj.2018.59.2.317