Osteoporos Int (2009) 20:1233–1240

DOI 10.1007/s00198-008-0794-9

ORIGINAL ARTICLE

Preventative effect of exercise against falls in the elderly:

a randomized controlled trial
J. Iwamoto & H. Suzuki & K. Tanaka & T. Kumakubo &
H. Hirabayashi & Y. Miyazaki & Y. Sato & T. Takeda &
H. Matsumoto

Received: 20 January 2008 / Accepted: 19 September 2008 / Published online: 15 November 2008
# International Osteoporosis Foundation and National Osteoporosis Foundation 2008

Abstract groups. The daily exercise, which consisted of calisthenics,

Summary The present study was conducted to determine the body balance training (tandem standing, tandem gait, and
effect of 5-month exercise program on the prevention of falls unipedal standing), muscle power training (chair-rising
in the elderly. The exercise training, which consisted of training), and walking ability training (stepping), were
calisthenics, body balance training, muscle power training, performed 3 days/week only in the exercise group. No
and walking ability training 3 days/week improved the exercise was performed in the control group.
indices of the flexibility, body balance, muscle power, and Results After the 5-month exercise program, the indices of
walking ability and reduced the incidence of falls compared the flexibility, body balance, muscle power, and walking
with non-exercise controls. The present study showed the ability significantly improved in the exercise group com-
beneficial effect of the exercise program aimed at improving pared with the control group. The incidence of falls was
flexibility, body balance, muscle power, and walking ability significantly lower in the exercise group than in the control
in preventing falls in the elderly. group (0.0% vs. 12.1%, P=0.0363). The exercise program
Introduction The present study was conducted to determine was safe and well tolerated in the elderly.
the effect of exercise on the prevention of falls in the Conclusions The present study showed the beneficial effect
elderly. of the exercise program aimed at improving flexibility,
Methods Sixty-eight elderly ambulatory volunteers were body balance, muscle power, and walking ability in
randomly divided into two groups: the exercise and control preventing falls in the elderly.

J. Iwamoto (*) : T. Takeda : H. Matsumoto T. Kumakubo

Department of Sports Medicine, Kumakubo Orthopaedic Clinic,
Keio University School of Medicine, Nakano-ku,
35 Shinanomachi, Shinjuku-ku, Tokyo, Japan
Tokyo 160-8582, Japan
e-mail: [email protected] H. Hirabayashi
Department of Orthopaedic Surgery,
J. Iwamoto Tokyo Adventist Hospital,
Department of Orthopaedic Surgery, Keiyu Orthopaedic Hospital, Suginami-ku,
Gunma, Japan Tokyo, Japan

H. Suzuki Y. Miyazaki
Department of Orthopaedic Surgery, Kawakita General Hospital, Miyazaki Orthopaedic Clinic,
Suginami-ku, Setagaya-ku,
Tokyo, Japan Tokyo, Japan

K. Tanaka Y. Sato
Kei Medical Clinic, Department of Neurology, Mitate Hospital,
Nakano-ku, Tagawa,
Tokyo, Japan Fukuoka, Japan
1234 Osteoporos Int (2009) 20:1233–1240

Keywords Body balance . Exercise . Fall . Muscle power . Surgery of two hospitals and three Orthopaedic Clinics in
Walking ability Suginami, Nakano, and Setagaya Wards, Tokyo, Japan
between July 2006 and March 2007 were recruited to our
trial. The inclusion criteria were an age of more than
Introduction 50 years, fully ambulatory, and being able to measure
parameters as described below, and the exclusion criteria
Fall-related injuries, including head injuries and fractures, were severe gait disturbance with some aids, severe round
are serious problems in the elderly, as they often lead to back due to osteoporotic vertebral fractures, acute phase of
prolonged or even permanent disability. Thus, prevention of diseases, and severe cardiovascular disease. The mean age
falls and therefore of the injuries associated with them of the participants was 76.4 years (range, 66–88 years). The
would reduce disability, improve the quality of life, and physical activity level at baseline was considered compar-
reduce the costs of health care. Impairment of muscle atively low in all of the participants because none of them
strength and muscle power of the lower extremities, had been laborers or had been engaged in any regular or
balance/postural control, and walking ability has been leisure time sporting activities.
recognized as important risk factors for falls [1]. These We assessed the ratio of male to female subjects, age, body
parameters are known to become progressively more weight, height, body mass index, history of falls in the past
impaired with aging [2], suggesting the increased risk for 3 months and fractures after 50 years of age, and the indices of
falls in the elderly. flexibility (finger floor distance [FFD] with the body flexed in
Muscle strength should be distinguished from muscle the anterior, right, and left directions), body balance (tandem
power; muscle strength is defined as the maximal force that standing time, tandem gait step number, and unipedal standing
a muscle can produce against a give resistance, while time), muscle power (timed up and go [TUG] [10] and chair-
muscle power is defined as the product of force and speed rising time [five times]), and walking ability (10-m walking
[1, 3]. The former is related to bone strength, whereas the time and walking step length). Tandem standing time and
latter is related to falling [1, 3–5]. Thus, improvement of unipedal standing time were determined by taking the mean
muscle power rather than muscle strength would appear to values of the right and left sides.
be important in the prevention of falls in the elderly. The subjects were randomly divided into two groups: the
Exercise is generally accepted to be effective for the exercise and control groups (n=34 in each group). Main
prevention of falls in the elderly. A meta-analysis study has diseases that could possibly affect physical activity in the
demonstrated that exercise is effective for lowering the risk of participants were osteoporosis with or without radiographic
falls in the elderly and that the consequent reduction in the vertebral fractures (n=13), sciatica due to lumbar spinal
incidence of fall-related injuries reduces health care costs [6]. canal stenosis (n = 8), knee osteoarthritis (n = 7), and
Furthermore, a systematic review has demonstrated that spondylosis (n=6) in the exercise group, and osteoporosis
muscle strengthening, balance exercises, and regular practice with or without radiographic vertebral fractures (n=11),
of the internal martial art therapy of tai chi chuan are sciatica due to lumbar spinal canal stenosis (n = 8),
effective for preventing fractures in the elderly [7–9]. spondylosis (n=8), knee osteoarthritis (n=5), hip osteoar-
Theoretically, however, improvement of muscle power of thritis (n = 1), and elbow arthritis (n = 1). All of the
the lower extremities, balance/postural control, and walking degenerative diseases were mild to moderate. Thus, the
ability by exercise is considered to be important in the healthiness and subsequent medication of the participants
prevention of falls in the elderly. In particular, Runge et al. [1] were similar at baseline in the two groups. The daily
reported that whole-body vibration exercise improved chair- exercise program (Table 1) consisted of calisthenics, body
rising time in terms of muscle power in the elderly. With the balance training (tandem standing, tandem gait, and unipedal
exception of this study, however, less attention would appear standing), muscle power training (chair-rising training), and
to have been paid to muscle power in the literature as far as walking ability training (stepping). All of the exercises were
the prevention of falls is concerned. The present study was supervised and performed in the clinics or hospitals 3 days/
conducted to determine the effect of exercise on the week only in the exercise group by taking about 30 min. So,
prevention of falls in the elderly, aimed at improving the compliance with the exercises was 100%. No exercise
flexibility, body balance, muscle power, and walking ability. was undertaken in the control group. The period of this study
was 5 months. The incidence of fall and fracture as well as
the above-mentioned indices of the flexibility, body balance,
Methods muscle power, and walking ability was assessed 2.5 and
5 months after the start of the trial. In particular, information
Sixty-eight elderly ambulatory volunteers (seven men and regarding falls and fractures was obtained every week by
61 women) who visited the Department of Orthopaedic directly asking the participants.
Osteoporos Int (2009) 20:1233–1240 1235

Table 1 Daily exercise program in the exercise group the ITT analysis at the baseline and 2.5 months after the
Calisthenics start of the trial, dropping to 33 subjects at 5 months after
Anterior, right, left, and posterior flexion of the body: 5 times in each the start of the trial.
direction a day
Body balance training Anthropometry and baseline physical function of the study
Tandem standing (3 min in each leg forward: 2 sets a day) subjects
Tandem gait (10 steps: 5 sets a day)
Unipedal standing (3 min in each leg: 2 sets a day)
Tables 2 and 3 show the anthropometry and baseline
Muscle power training
Chair-rising training (10 times: 3 sets a day)
physical function of the study subjects, respectively. The
Walking ability training (stepping) mean age was significantly different between the two
Having one step in the forward, back, right, and left directions: 10 groups (74.6 years in the exercise group and 78.2 years in
times for each stepping a day the control group, P<0.01). However, there were no
significant differences in any other baseline characteristics
All of the exercises were supervised and performed in the clinics or
hospitals 3 days/week only in the exercise group by taking about
including the ratio of male to female subjects, body weight,
30 min height, body mass index, and percentage of subjects who
had experienced falls in the past 3 months and fractures
after 50 years of age. There were ten fallers (29.4%) in the
Informed consent was obtained from each of the subjects exercise group at baseline: five had experienced only one
prior to their participation in the study. The protocol was fall, and five had experienced two falls. There were also ten
approved by the Ethical Committee of Keiyu Orthopaedic fallers (29.4%) in the control group at baseline: four had
Hospital. experienced only one fall, five had experienced two falls,
An intention-to-treat (ITT) analysis was adopted. Data and one had experienced four falls. There was a history of
are expressed as means±standard deviation in tables and 14 clinical fractures in 11 participants (32.4%) of the
means ±95% confidence intervals (CIs) in figures. The use exercise group: eight vertebral fractures, three hand
of 95% CIs facilitates the distinction between statistical fractures, two rib fractures, and one forearm fracture. There
significance and clinical significance or practical impor- also was a history of 12 clinical fractures in eight
tance in figures. The Mann–Whitney U test was used to participants (23.5%) of the control group: six vertebral
compare baseline characteristics between the two groups. A fractures, two forearm fractures, two foot fractures, one rib
one-way analysis of variance (ANOVA) with repeated fracture, and one ankle fracture. There were no significant
measurements was used to analyze the longitudinal changes difference in any of baseline physical function indices of
in physical function parameters within a group. An analysis flexibility (FFD), body balance (unipedal standing time,
of covariance (ANCOVA) was used to compare changes in tandem gait step number, tandem standing time), muscle
physical function parameters at each time point between the power (TUG, chair-rising time), and walking ability (10-m
two groups using baseline values as covariates. The chi- walking time, step length) between the two groups.
square test was used to compare the baseline characteristics
such as the ratio of male to female subjects and the incidence
of falls and fractures between the two groups. All statistical
analyses were performed using the Stat View J-5.0 program Table 2 Anthropometry of the study subjects
(SAS Institute, Cary, NC, USA). The significance level was
Exercise Control P value
set at P<0.05 for all the comparisons.
N 34 34
Male/Female 5/29 2/32 NS
Results Age (years) 74.6±5.6 78.2±5.6 <0.01
Height (m) 1.52±0.05 1.52±0.05 NS
Number of subjects who were included in the ITT analysis Body weight (kg) 52.1±11.0 48.0±7.9 NS
Body mass index (kg/m2) 22.4±4.1 20.7±1.9 NS
Faller in the past 3 months (%) 29.4 29.4 NS
All participants in the exercise group completed the 5-
History of clinical fracture (%) 32.4 23.5 NS
month trial. However, one participant in the control group
dropped out from the trial because of noncompliance at Data are expressed as means±standard deviation. The Mann–Whitney
5 months after the start of the trial. Thus, 34 subjects in the U test was used to compare the anthropometry between the two
groups. The chi-square test was used to compare the baseline
exercise group were included in the ITT analysis at the characteristics such as the ratio of male to female subjects and the
baseline and 2.5 and 5 months after the start of the trial, incidence of falls and fractures between the two groups
whereas 34 subjects in the control group were included in NS not significant
1236 Osteoporos Int (2009) 20:1233–1240

Table 3 Physical function of the study subjects—flexibility, body time), muscle power (TUG, chair-rising time), and walking
balance, muscle power, and walking ability indices
ability (10-m walking time, step length), respectively. The
Exercise Control P value one-way ANOVA with repeated measurements showed that
all indices of flexibility, body balance, muscle power, and
FFD Anterior flexion (cm) 10.0±6.7 10.1±9.0 NS walking ability significantly improved in the exercise
Right flexion (cm) 38.1±6.8 37.8±10.3 NS
group, whereas FFD in the right and left directions
Left flexion (cm) 39.1±8.2 39.7±8.4 NS
Unipedal standing time (s) 39.8±25.0 28.2±22.7 NS significantly worsened in the control group. The ANCOVA
Tandem gait step number 18.6±20.2 12.9±17.7 NS showed that, after the 5-month intervention, there were
Tandem standing time (s) 80.5±48.5 65.3±56.1 NS significant differences in FFD in the right and left
Timed up and go (s) 5.2±2.4 7.3±7.6 NS directions, unipedal standing time, tandem gait step
Chair-rising time (s) 14.9±5.5 18.0±11.8 NS number, tandem standing time, TUG, chair-rising time,
10-m walking time (s) 10.0±4.4 10.8±4.0 NS 10-m walking time, and step length between the exercise
Step length (cm) 62.2±11.3 59.9±15.4 NS
and control groups. However, FFD in the anterior direction
Data are expressed as means±standard deviation. The Mann–Whitney did not differ significantly between the two groups.
U test was used to compare the baseline characteristics between the
two groups Effect of exercise on the incidence of falls and fractures
FFD finger floor distance, NS not significant.

Four participants in the control group experienced one fall

Effect of exercise on the indices of flexibility, body balance, each during the 5 months intervention period. Of four falls,
muscle power, and walking ability one was due to a stumble of the toe, and three were caused
by lurches. There was no multiple faller during the 5 months
Figures 1, 2, 3, and 4 show the effect of exercise on the intervention period. In particular, a participant of the
indices of flexibility (FFD), body balance (unipedal control group who had experienced four falls in the past
standing time, tandem gait step number, tandem standing 3 months at baseline (multiple faller) had no fall during the

Fig. 1 Finger floor distance with anterior, right, and left flexion of the 0.05). The analysis of covariance showed that, after the 5-month
body. Data are expressed as means ±95% confidence intervals. The intervention, there were significant differences in FFD in the right and
one-way analysis of variance with repeated measurements showed that left directions between the exercise and control groups. However, FFD
FFD in the anterior direction significantly improved in the exercise in the anterior direction did not differ significantly between the two
group (P<0.01) but not in the control group and that FFD in the right groups. NS not significant, asterisks significant changes by the one-
and left directions significantly improved in the exercise group (both way ANOVA with repeated measurements
P<0.05) but significantly worsened in the control group (both P<
Osteoporos Int (2009) 20:1233–1240 1237

Fig. 2 Unipedal standing time, tandem gait step number, and tandem covariance showed that, after the 5-month intervention, there were
standing time. Data are expressed as means ±95% confidence significant differences in unipedal standing time, tandem gait step
intervals. The one-way analysis of variance with repeated measure- number, and tandem standing time between the exercise and control
ments showed that unipedal standing time, tandem gait step number, groups. NS not significant, asterisks significant changes by the one-
and tandem standing time significantly improved in the exercise group way ANOVA with repeated measurements
(all P< 0.0001) but not in the control group. The analysis of

Fig. 3 Timed up and go and chair-rising time. Data are expressed as 5-month intervention, there were significant differences in TUG and
means ±95% confidence intervals. The one-way analysis of variance chair-rising time between the exercise and control groups. NS not
with repeated measurements showed that TUG and chair-rising time significant, asterisks significant changes by the one-way ANOVA with
significantly improved in the exercise group (both P<0.0001) but not repeated measurements
in the control group. The analysis of covariance showed that, after the
1238 Osteoporos Int (2009) 20:1233–1240

Fig. 4 Ten-meter walking time and step length. Data are expressed as showed that, after the 5-month intervention, there were significant
means ±95% confidence intervals. The one-way analysis of variance differences in 10-m walking time and step length between the exercise
with repeated measurements showed that 10-m walking time and step and control groups. NS not significant, asterisks significant changes by
length significantly improved in the exercise group but not in the control the one-way ANOVA with repeated measurements
group (P<0.01 and P<0.05, respectively). The analysis of covariance

5 months intervention period. The incidence of falls during body balance, muscle power, and waling ability, leading to a
the study period was significantly lower in the exercise reduced incidence of falls.
group than in the control group (0.0% vs. 12.1%, χ2 = A consensus has been reached with respect to exercise
4.383, P=0.0363). Above four falls resulted in bruises or programs for the elderly; that is, a combination of muscle-
sprains of the upper extremities, which required no strengthening exercises of the back and lower extremities,
intensive treatment and healed within several days. There balance exercises, and walking may be effective to prevent
were no fall-related fractures reported in either group. vertebral and nonvertebral fractures [11]. Theoretically,
however, improvement of muscle power of the lower
Adverse events extremities, balance/postural control, and walking ability
are all important to prevent falls in the elderly. Although
During the study period, no serious adverse events, such as muscle strength, which is related to bone strength, has often
severe fall-related injuries or adverse cardiovascular effects, been assessed in the literature, less attention has been paid
were observed. to muscle power despite the fact that it is related to falling.
In the present study, we therefore included the indices, not
of muscle strength, but rather of muscle power such as the
Discussion chair-rising time and TUG.
However, we did not apply brisk walking exercises to
The muscle power of the lower extremities, balance/postural improve walking speed because there are controversial
control, and walking ability are important factors in the reports on the effect of walking exercise on the incidence of
prevention of falls in the elderly. The focus of the present falls. Feskanich et al. [12] showed in a cohort study that,
study was (1) whether the exercise program (three times per among postmenopausal women, walking for at least 4 h/
week, aimed at improving flexibility, body balance, muscle week was associated with a 41% lower risk of hip fracture
power, and walking ability) would improve those related compared with less than 1 h/week. On the other hand,
parameters in the elderly; and (2) whether improvement of Gillespie et al. [7] showed in a systematic review that brisk
these parameters, if any, would be useful to prevent falls. walking increased the risk of upper limb fractures in elderly
Thus, a randomized controlled trial was designed. We women. Based on the hierarchy of the evidence, we believe
confirmed that our exercise program improved flexibility, more in the result of the systematic review. Therefore, we
Osteoporos Int (2009) 20:1233–1240 1239

applied only stepping, which might be safe from the point There were strengths in this study. First, this randomized
of view that exercise therapy should be safe. However, 10- controlled trial was strictly performed not by exercise-
m walking time and TUG were significantly improved in related experts but mainly by general practitioners so that
the exercise group compared with the control group, our exercise program could be performed by general
suggesting the usefulness of stepping in the forward, back, practitioners without using any special machines. Second,
right, and left directions in improving walking speed in the exercises aimed at improving flexibility, body balance,
elderly. muscle power, and walking ability improved the parameters
Exercise also increased the walking step length. Basically, related to these functions, leading to a reduction in the
each stride during walking consists of the stance and swing incidence of falls as a primary end point. Third, the exercise
phases. Thus, increased unipedal standing time can produce program was safe and well tolerated in the elderly. These
more stable walking. That is, the more the stance phase of strengths suggest the usefulness and convenience of our
each leg was stabilized by exercise, the greater the swing of exercise program in the prevention of falls in the elderly.
the other leg becomes, resulting in an increase in step length. The limitations of this study should also be discussed.
Because impaired walking ability is associated with decreased First, the study period was short (5 months). It is a
walking step length [13], increased walking step length may recognized fact that long-term exercise is needed to reduce
also indicate improved walking ability. Because it is the life-time risk of falls and fall-related injuries in the
important for the elderly to increase walking step length to elderly. However, because our exercise program proved
touch the ground with the heel during walking in order to easy for our elderly subjects to continue without any
prevent falls caused by toe-contact-related stumbling, we difficulty, we believe that it could be continued under the
believe that improvement of walking step length could also instruction of general practitioners. Second, the number of
lead to a reduction in the fall-related risk. the study subjects was small, even though the statistically
We applied exercise 3 days/week. The intensity and significant results were obtained. Thus, further studies are
frequency of the exercise program were considered to be needed to resolve the limitations.
reasonable for the elderly (mean age, 76.4 years) to be In conclusion, the present study showed the beneficial
continued without any fatigue and difficulty for 5 months. effect of an exercise program aimed at improving flexibility,
Exercise was not only effective to prevent falls but also was body balance, muscle power, and walking ability to prevent
well tolerated, and no serious adverse events, such as fall- falls in the elderly. Our exercise program improved flexibil-
related injuries or adverse cardiovascular effects, were ity (FFD with the body flexed in the right and left directions),
observed in any of the subjects during the exercise body balance (tandem standing time, tandem gait step
program, suggesting the safety of our exercise program. number, and unipedal standing time), muscle power (TUG
Flexibility in terms of FFD in the right and left directions and chair-rising time), and walking ability (10-m walking
was impaired during the 5-month period in the control group time and step length), leading to a reduced incidence of falls,
(only 3–4 cm increases in the FFD). The reason for this and the beneficial effect of our exercise program in
remains uncertain because there was nobody with a significant preventing falls in the elderly was thus confirmed. Further-
collapse in his/her status or who had a significant change in more, the exercise program was safe and well tolerated by all
medication. However, this result possibly suggests that not our subjects.
only muscle strength and muscle power of the lower
extremities, balance/postural control, and walking ability [2] Acknowledgments We would like to thank Mr. Fusamitsu Inui,
Mr. Toshihiro Yamaguchi, and other members of the Tokyo Branch
but also flexibility could be impaired with aging. So, an of Banyu Co. Ltd., Tokyo, Japan for their assistance in the
exercise regimen aimed at improving flexibility should be preparation of the trial.
included in the exercise therapy.
There is a criticism that the subjects of the control group
Conflicts of interest None.
were older and tended to be frailer than those of the
exercise group despite the similar proportion of fallers in
the two groups and less history of clinical fractures in the
References
control group at baseline. Lacking in statistically significant
differences in physical function parameters at baseline
1. Runge M, Rehfeld G, Resnicek E (2001) Balance training and
between the two groups might not be due to similarity of exercise in geriatric patients. J Musculoskel Neuronal Interact
the parameters but be attributable to a large variation of the 1:61–65
parameters. In order to resolve this issue, ANCOVA was 2. Asmussen E (1980) Aging and exercise. Environ Physiol 3:419–
428
used to compare changes in physical function parameters at
3. Runge M, Hunter G (2006) Determinants of musculoskeletal
each time point between the two groups using baseline frailty and the risk of falls in old age. J Musculoskel Neuronal
values as covariates. Interact 6:167–173
1240 Osteoporos Int (2009) 20:1233–1240

4. Frost HM (1997) Defining osteopenias and osteoporosis: another of Tai Chi and computerized balance training. J Am Geriatr Soc
view (with insights from a new paradigm). Bone 20:385–191 44:489–497
5. Schiessl H, Frost HM, Jee WSS (1998) Estrogen and bone-muscle 10. Podsiadlo D, Richardson S (1991) The timed “up & go:” A test of
strength and mass relationship. Bone 22:1–6 basic functional mobility for frail elderly persons. J Am Geriatr
6. Gardner MM, Robertson MC, Campbell AJ (2000) Exercise in Soc 39:142–148
preventing falls and fall related injuries in older people: a review 11. Allen SH (1994) Exercise considerations for postmenopausal
of randomized controlled trials. Br J Sports Med 34:7–17 women with osteoporosis. Arthritis Care Res 7:205–214
7. Gillespie LD, Gillespie WJ, Robertson MC, Lamb SE, Cumming 12. Feskanich D, Willett W, Colditz G (2002) Walking and leisure-
RG, Rowe BH (2001) Interventions for preventing falls in elderly time activity and risk of hip fracture in postmenopausal women.
people. Cochrane Database Syst Rev 3: CD000340 JAMA 288:2300–2306
8. Lane JM, Nydick M (1999) Osteoporosis: current modes of 13. Hoshino K, Beppu M, Ishii S, Masuda T, Hibino Y, Oyake Y,
prevention and treatment. J Am Acad Orthop Surg 7:19–31 Aoki H, Sudou K, Iida Y (2002) The gait analysis of the elderly at
9. Wolf SL, Barnhart HX, Kutner NG, McNeely E, Coogler C, Xu T the fall prevention exercise class. J Physical Medicine 13:113–117
(1996) Reducing frailty and falls in older persons: an investigation (in Japanese)