A Randomized Controlled Trial of Weight-Bearing Versus

NonWeight-Bearing Exercise for Improving Physical Ability

After Usual Care for Hip Fracture
Catherine Sherrington, PhD, Stephen R. Lord, PhD, Robert D. Herbert, PhD
ABSTRACT. Sherrington C, Lord SR, Herbert RD. A
randomized controlled trial of weight-bearing versus non
weight-bearing exercise for improving physical ability after
usual care for hip fracture. Arch Phys Med Rehabil 2004;85:
710-6.
Objective: To compare the effects of weight-bearing and
nonweight-bearing home exercise programs and a control
program on physical ability (strength, balance, gait, functional
performance) in older people who have had a hip fracture.
Design: Randomized controlled trial with 4-month follow-up.
Setting: Australian community-dwellers (82%) and resi-
dents of aged care facilities who had completed usual care after
a fall-related hip fracture.
Participants: One hundred twenty older people entered the
trial, 40 per group (average age standard deviation, 799y)
and 90% completed the 4-month retest.
Intervention: Home exercise prescribed by a physical ther-
apist.
Main Outcome Measures: Strength, balance, gait, and
functional performance.
Results: At the 4-month retest, there were differences be-
tween the groups in the extent of improvement since the initial
assessment for balance (F
10,196
2.82, P.001) and functional
performance (F
6,200
3.57, P.001), but not for strength
(F
12,190
1.09, P.37) or gait (F
8,200
.39, P.92). The
weight-bearing exercise group showed the greatest improve-
ments in measures of balance and functional performance
(between-group differences of 30%40% of initial values).
Conclusions: A weight-bearing home exercise program can
improve balance and functional ability to a greater extent than a
nonweight-bearing program or no intervention among older peo-
ple who have completed usual care after a fall-related hip fracture.
Key Words: Exercise therapy; Hip fractures; Physical
therapy; Rehabilitation.
2004 by the American Congress of Rehabilitation Medi-
cine and the American Academy of Physical Medicine and
Rehabilitation
F
ALLS AND FRACTURES in older people are of major
public health signicance. One in 3 older people experi-
ence 1 or more falls each year
1
and about 1 in six 50-year-old
Western women will have at least 1 hip fracture before they
die.
2
It is estimated that by 2050 the total global cost of hip
fractures will be $131.5 billion annually.
3
Unfortunately, out-
comes after hip fracture are often poor.
4
Despite such poor outcomes, few clinical trials have inves-
tigated medium to long-term rehabilitation strategies after hip
fracture. In a study of home exercise among 42 people 7
months after hip fracture,
5
we found greater improvements in
quadriceps strength, weight-bearing ability, and walking speed
in those who completed a daily weight-bearing home exercise
program than in a no-intervention control group. However,
another study, by Tinetti et al,
6
found a program of systematic
home-based physical and functional therapy
7
to be no more
effective than the usual home-based rehabilitation intervention
after hip fracture.
In addition to the paucity of research into the subject, there
is also a lack of consensus among researchers and clinicians
about which exercise regimen is best for older people after hip
fracture. Traditionally, nonweight-bearing bed exercises have
been prescribed.
8
However, it has been argued that rehabilita-
tion exercise is most effective when it is closely related to the
particular task that it is intended to improve (eg, standing up,
walking, stair climbing).
9
This argument is based on evidence
that the training response is specic to the method of training
employed.
10
Exercise conducted in a weight-bearing posture
(ie, standing) would be more relevant to daily tasks.
To address these issues, we conducted a randomized con-
trolled trial that compared the effects of weight-bearing and
nonweight-bearing home exercise programs and those of a
no-intervention control program among older people who had
completed usual care after a fall-related hip fracture.
METHODS
Participants
One hundred twenty older people participated in the study.
The average age standard deviation (SD) of subjects was
799 years (range, 5795y). Eighty-two percent of the sample
lived in the community and the remainder lived in low- or
high-care residential aged care facilities. Ninety-six subjects
(80%) were women.
Subjects were recruited through 6 hospitals in Sydney, Aus-
tralia. Depending on the hospital, subjects were identied dur-
ing visits to orthopedic and rehabilitation wards, from lists
provided by medical records departments, and through infor-
mation provided by ward staff. Subjects were excluded if they
were unable to complete the assessments or the home-exercise
program because of (1) severe cognitive impairment, (2) med-
ical conditions, or (3) complications from the fracture resulting
in delayed healing and associated weight-bearing restrictions.
Subjects who did not meet the inclusion criteria were identied
through face-to-face or telephone interviews with them and/or
From the Prince of Wales Medical Research Institute, University of New South
Wales, Sydney (Sherrington, Lord); and School of Physical Therapy, University of
Sydney, Sydney (Herbert), Australia.
Supported by the Health Research Foundation Sydney South West, Arthritis
Foundation of Australia, and National Health and Medical Research Council Part-
nership in Injury Grant.
Presented in part at the Australian Physical Therapy Associations 7th International
Physical Therapy Congress, May 2002, Sydney, Australia.
No commercial party having a direct nancial interest in the results of the research
supporting this article has or will confer a benet upon the author(s) or upon any
organization with which the author(s) is/are associated.
Correspondence to Catherine Sherrington, PhD, Prince of Wales Medical Research
Institute, University of New South Wales, Barker St, Randwick, Sydney, NSW 2031,
Australia, e-mail: c.sherrington@unsw.edu.au. Reprints are not available from the
author.
0003-9993/04/8505-7975$30.00/0
doi:10.1016/S0003-9993(03)00620-8
710
Arch Phys Med Rehabil Vol 85, May 2004
their caregivers and by information provided by hospital and/or
aged care facility staff.
Ethics approval was obtained from the South Western Syd-
ney Area Health Service Research Ethics Committee, the Hope
Healthcare Ethics Committee, and the Central Sydney Area
Health Service Ethics Review Committee. Informed consent
was obtained from all subjects before their participation.
Assessment Procedure
Structured interviews and physical assessments were done in
subjects homes before the intervention, and 1 month and 4
months after the initial assessment. The rst author and 2 other
physical therapists conducted assessments. Assessors were not
blinded to group allocation. With few exceptions, the same
assessor conducted all 3 assessments for a particular subject.
To maximize standardization between testers, several training
sessions were held and a measurement protocol was followed.
Footwear was standardized across the 3 assessments. Each
assessment took approximately 45 minutes.
Measurement Tools
Strength. The maximal voluntary strength of the knee
extensor muscles was tested using a strap with a spring
gauge.
11
Subjects sat on a tall chair with a strap around the leg
10cm above the ankle joint. The hip and knee joints were at 90
angles. In 2 trials, the subject attempted to pull against the strap
assembly with maximal force for 2 to 3 seconds, and the
greater force was recorded.
A hand-held dynamometer
12
(positioned 5cm proximal to
the malleoli) was used to measure isometric force generation of
the hip abductor and hip exor muscles. Subjects lay supine on
a plinth with the hips in the anatomic position. Again, 2
attempts were made for each test and the higher value recorded.
Lateral step-up ability was measured with the subject stand-
ing. With both feet adjacent, the subject placed 1 foot onto a
10-cm block and attempted to lift the other leg off the ground.
The subjects need for support was documented on a 5-point
scale (unable, required assistance from another person, re-
quired 2 hand supports, required 1 hand support, required no
hand support). Although this procedure has previously been
described as part of an exercise program
5
and is used clinically
to evaluate patient performance, its use as a measurement tool
has yet to be investigated.
Balance. Postural sway was measured with the subject
standing on the oor and on a 7-cm medium-density foam
rubber mat, using a portable sway meter as described else-
where.
11
Total sway path over 30 seconds was measured in
millimeters.
Functional reach,
13
the distance a subject can reach forward
without moving the feet, was measured in centimeters using a
retractable tape measure. Two attempts were completed and the
better effort was recorded.
The step test,
14
which measures the number of times that a
subject can step up onto a 5-cm block without hand support in
15 seconds, was assessed for each leg. For this test, the subject
stood in front of the block and placed the whole of 1 foot up
onto and then down off the block repeatedly.
Gait. Measures were rst taken while the subjects walked
6m at their usual comfortable pace with the least supportive aid
that the examiner judged to be safe. Time taken was recorded
with a stopwatch. Walking aid used and the number of steps
taken were also recorded. Subjects then walked 6m as quickly
as they could, and the same measures were repeated.
Functional performance. The Physical Performance and
Mobility Examination
15
(PPME) involves measures of bed
mobility, transfer skills, multiple stands from a chair, standing
balance, step-up ability, and ambulation. Assistance and time
taken for each category are quantied, and each is scored on a
3-point scale (high pass, 2; low pass, 1; fail, 0). The maximum
possible score is 12. The time taken to stand up from and sit
down on a 45-cm chair 5 times and to move from lying to
sitting over the side of a bed were also recorded.
Self-report. The structured interview obtained informa-
tion on the following measures: self-rated fall risk, balance,
health, quality of night-time sleep, pain, mobility and activity
levels, and impairments relating to activities of daily living
(ADLs).
Intervention
The study design involved 3 groups, a weight-bearing home
exercise (WBE) group, a nonweight-bearing home exercise
(NWBE) group, and a control group. At their rst appoint-
ments, subjects were allocated to groups using assignments
sealed in opaque envelopes. The randomization schedule was
produced with a random number table, with subjects being
randomized to groups in blocks of 6.
The WBE group did its exercises in weight-bearing posi-
tions. This program has been implemented in several Sydney
hospitals, where it has been modied from programs used for
rehabilitation after stroke.
9,16
The exercises were: sit-to-stand
(repeated stands from a chair or adjustable-height exercise
plinth), lateral step-up (as described in the testing procedure),
forward step-up-and-over (stepping onto a block with both legs
and down off it again), forward foot taps (tapping 1 foot up
onto a block while supporting the weight on the other leg), and
a stepping grid (stepping in different directions as guided by
marks on the oor). These exercises were initially conducted
with tables, chairs, or walking aids used for support. The
exercises were progressed by increasing the number of repeti-
tions, lessening the hand support, increasing the height of the
blocks, or decreasing the chair height.
The NWBE group carried out all exercises in a nonweight-
bearing (supine) position as commonly prescribed after hip
fracture.
8,17
These exercises were: hip abduction (sliding the
straight leg out to the side), hip exion (lifting the straight leg),
hip and knee exion and extension (sliding the heel toward the
buttock by bending the hip and knee), end of range knee
extension (straightening the bent knee over a wedge), and ankle
dorsiexion and plantarexion. The exercises were progressed
by increasing the number of repetitions.
For the exercise groups, a physical therapist prescribed the
exercise program based on predetermined guidelines, after
conducting the initial assessment. One to 5 of the 5 exercises
were chosen and the number of repetitions suggested was based
on the subjects performance of the exercises. Subjects were
provided with line drawings of the exercises and asked to
complete the program daily and to record the number of rep-
etitions completed. Participants in the WBE group were pro-
vided with 1 or 2 blocks (5-10cm thick) to use for the step-up
exercises. A follow-up visit was made 1 week later to assess
performance of the exercises, to provide feedback on perfor-
mance, and to revise the suggested repetitions. At 1- and
4-month follow-up assessment visits, exercise performance
was also assessed, more exercises were prescribed, and the
number of repetitions was increased if the subjects felt that they
could do more.
The control group did not receive any interventions.
Statistical Analysis
Data were analyzed with multivariate general linear models.
Each model assessed the effects of intervention on the differ-
711 EXERCISE AFTER HIP FRACTURE, Sherrington
Arch Phys Med Rehabil Vol 85, May 2004
ences between the preintervention assessment and 1- and
4-month assessment performance on each of the 4 multivariate
physical domains (strength, balance, gait, functional perfor-
mance). Multivariate analysis of variance (ANOVA) discrimi-
nant function coefcients were then computed to assess the
relative contribution of the domain variables to the signicant
results. Univariate analyses from the general linear model are
also reported here. If they revealed signicant between-group
differences (P.05), the Hochberg post hoc test
18
was used to
identify differences between pairs of groups.
Between-group comparisons were conducted for preinter-
vention means and postintervention measures not covered by
the multivariate analyses (ie, lateral step-up, walking aid use,
self-report measures, time between assessments) using factorial
(group time) ANOVA for continuous measures, the Mann-
Whitney U or Kruskal-Wallis tests for ordinal measures, and
chi-square tests for dichotomous variables.
If continuous data were found to be skewed, logs of the
scores were computed and statistical testing was carried out on
the more normally distributed variables. For several variables,
data for some subjects were unobtainable because of impaired
performance. Where this occurred, the mean 3 SDs (in the
direction of impaired performance) was used as the test value.
All available data were analyzed by initial group assignment
(ie, an intention-to-treat approach). The analyses were per-
formed using SPSS
a
for Windows.
RESULTS
Preintervention: Subject Characteristics
Subject characteristics are summarized in table 1. There
were no clinically important or statistically signicant differ-
ences between the 3 study groups at the initial assessment.
Subject Adherence to Protocols and Experience of
Exercise
All 120 subjects completed the initial assessment and 108
(90%) underwent at least part of the 1- and 4-month assess-
ments. Figure 1 shows the proportion of subjects lost to fol-
low-up at 4 months by group. In addition, 2 subjects at 1 month
and 4 subjects at 4 months were unable to complete the entire
physical assessment because of ill health. Subject numbers
completing each test are shown in tables 2 through 5.
Statistical testing did not reveal any signicant differences in
the time between assessments for the 3 groups. The mean
number of days between preintervention and 1-month assess-
ments was 33.817.0 for the WBE group, 31.97.0 for the
NWBE group, and 30.54.5 for the controls; and between the
preintervention and 4-month assessments it was 120.512.9
for the WBE group, 119.47.7 for the NWBE group, and
118.99.4 for the controls.
There were no statistically signicant differences in compli-
ance with the exercise program between the subjects in the 2
exercise groups who had follow-up assessments, as determined
by completion of exercise records and reports by the subjects.
At 1 month, 85% of the WBE group and 89% of the NWBE
group were doing the exercises at least 3 times weekly; at 4
months these gures were 69% and 73%. Data on actual
number of repetitions completed (from subject exercise logs)
were not analyzed because they were incomplete. There were
also no statistically signicant differences in subject reports of
pain from the exercises (at 1mo, 15% of the WBE group and
16% of the NWBE group reported moderate or marked pain; at
4mo, these gures were 29% and 14%, respectively) or their
perception of the usefulness of the exercises (at 1mo, 70% of
the WBE group and 66% of the NWBE group reported mod-
Table 1: Demographic, Health, Disability, and Mobility Characteristics of Intervention and Control Groups at Preintervention Assessment
Characteristics
Weight-Bearing Exercise
(n40)
NonWeight-Bearing Exercise
(n40)
Control
(n40)
Mean age (y) 80.17.5 79.18.9 77.28.9
Female 30 (75) 31 (78) 34 (85)
Community-dweller 31 (78) 34 (85) 33 (83)
Mean days since fracture 153.950.2 158.555.7 145.745.5
Left-sided fracture 19 (48) 24 (60) 23 (58)
Pain from fracture: mild or less 32 (80) 30 (75) 34 (85)
Mean medications taken 4.12.6 4.93.5 3.72.3
Mean illnesses reported 2.81.6 3.02.0 3.32.2
Health: excellent or very good 14 (35) 15 (38) 11 (28)
Balance: always steady 9 (23) 11 (28) 15 (38)
Fall risk: low 16 (40) 16 (40) 16 (40)
Falls in past 12mo: 1 17 (43) 8 (20) 14 (35)
Mental status: intact 28 (70) 31 (78) 31 (78)
Sleep quality: good or very good 21 (53) 21 (53) 15 (38)
Mean Katz ADLS score
23
3.22.0 3.12.4 3.22.5
Undertakes sport 1 (3) 4 (10) 3 (8)
Walks for exercise 17 (43) 18 (45) 20 (50)
Able to do heavy housework 10 (25) 11 (28) 13 (33)
Able to climb ight of stairs 23 (58) 28 (70) 24 (60)
Able to walk 800m 16 (40) 20 (50) 21 (53)
Hours on feet per day: 4 17 (43) 22 (55) 23 (58)
Indoor walking: unaided 14 (35) 20 (50) 19 (48)
Outdoor walking: unaided 4 (10) 9 (23) 5 (13)
Mean FAC score
24
4.40.9 4.60.6 4.30.7
NOTE. Values are mean SD or n (%).
Abbreviations: FAC, Functional Ambulation Categories; Katz ADLS, Katz Activities of Daily Living Scale.
712 EXERCISE AFTER HIP FRACTURE, Sherrington
Arch Phys Med Rehabil Vol 85, May 2004
erate or marked usefulness; at 4mo, these gures were 71% and
76%, respectively). There was a trend indicating a difference
between the groups for the perceived difculty of the exercises
at the 1-month assessment (15% of the WBE group and 8% of
the NWBE group found the exercises difcult or very difcult;
Mann-Whitney U429.5, z1.73, P.08), and a signicant
difference at the 4-month assessment (17% of the WBE group
and 0% of the NWBE group found the exercises difcult or
very difcult; Mann Whitney U363.0, z2.50, P.01).
Effect of the Intervention on Physical Performance
Tables 2 through 5 show preintervention and 1- and 4-month
assessment values for the continuous variables. Mean between-
group differences and their 95% condence intervals are re-
ported in the table footnotes.
At the 1-month assessment, no differences were found be-
tween groups in the extent of improvement since the initial
assessment for strength (F
12,198
.92, P.53), balance
(F
10,200
1.47, P.15), gait (F
8,204
.47, P.88), or functional
performance (F
6,204
2.03, P.06).
At the 4-month assessment, there were differences between
the groups in the extent of improvement since the initial as-
sessment for balance (F
10,196
2.82, P.001) and functional
performance (F
6,200
3.57, P.001), but not for strength
(F
12,190
1.09, P.37) or gait (F
8,200
.39, P.92). The stan-
dardized discriminant function coefcients for the variables
within the balance domain were .306 for the step test on the
affected leg, .449 for the step test on the nonaffected leg,
.662 for functional reach, .397 for sway distance on the
oor, and .092 for sway distance on foam. For the functional
performance domain, these values were .852 for timed sit-to-
stand, .273 for timed supine-to-sit, and .210 for PPME total
score.
There were also statistically and clinically signicant
between-group differences for several ordinal variables. A
larger proportion of people from the WBE group became
able to undertake the lateral step-up on the affected leg
without hand support at 1 month (WBE, 30%; NWBE, 11%;
control, 8%;
2
2
7.32, P.03) and on the nonaffected leg
at 4 months (WBE, 30%; NWBE, 11%; control; 8%;

2
2
10.15, P.006). At 4 months, more people in the WBE
group became able to walk unaided (WBE, 36%; NWBE,
6%; control, 14%,
2
2
20.02, P.001).
Effect of Interventions on the Self-Reported Measures
No differences between the groups were found for self-rated
fall risk, balance, health, quality of night-time sleep, pain,
activity levels, Katz Activities of Daily Living Scale, or Func-
tional Ambulation Categories performance at either 1- or
4-month assessments.
DISCUSSION
This study found between-group differences for the domains
of balance and functional ability but not for strength and gait,
with the weight-bearing exercise group showing the greatest
improvement. For the balance domain, the most important
contributors to this effect were improved functional reach and
step test abilities. This indicates that these subjects gained
better control of their center of mass while reaching and while
stepping, presumably because the exercises they undertook
provided a greater challenge to the postural control system. No
between-group differences were evident for improvements in
postural sway, which has been shown to be an important risk
factor for falling.
19
However, performance on this test is vari-
able, so improvements may be more difcult to demonstrate
than on other balance measures used. Previous exercise pro-
grams that have led to improved postural sway have been
supervised and of longer duration.
20
This program involved
few visits to subjects and was also conservative in its chal-
lenges to the postural control system. Because the exercises
Fig 1. Flow of subjects through the study.
713 EXERCISE AFTER HIP FRACTURE, Sherrington
Arch Phys Med Rehabil Vol 85, May 2004
were carried out at home, often while the person was alone,
safety was a prime consideration in exercise prescription.
The greatest contribution to the between-group difference in
functional abilities resulted from improved sit-to-stand perfor-
mance. The weight-bearing group improved the most on timed
sit-to-stand. This indicates the potential of this exercise pro-
gram to improve functional abilities that may have implications
for an older persons ability to continue to live independently
in the community. Each subcomponent of the PPME is mea-
sured on a 3-point scale, therefore, the lack of marked changes
in the PPME score may indicate the poor sensitivity of the scale
to smaller changes.
Multivariate analysis of the strength domain did not reveal
signicant between-group differences. However, the WBE
group did show a greater improvement in knee extension
strength and lateral step-up ability. Greater improvements in
muscle strength may have resulted from programs that specif-
ically targeted strength by overloading muscles in a more
intense and structured manner. There was also a correlation
between improvements in sit-to-stand ability and in knee ex-
tension strength for both the affected (r.28, P.005) and
nonaffected (r.23, P.02) leg.
Although the multivariate analysis of the gait domain found
no signicant between-group differences, when compared with
the other groups, more people in the WBE group were able to
walk unaided at the 4-month retest. This suggests better walking
ability and greater condence while walking, despite no change in
gait speed or stepping rate. It is possible that greater between-
group differences in these gait parameters could have resulted if
the WBE program had specically targeted these factors. A more
extensive WBE program could also include activities to prepare
participants to deal with real-life environmental challenges, such
as practice on ramps, stairs, and various surfaces.
The lack of between-group differences in the extent of
improvement in self-reported measures of functional ability,
general health, or activity levels indicates that the between-
group differences discussed above were not perceived to have
resulted in measurable effects on ability to perform everyday
activities. This may reect that self-reported functional out-
comes may be difcult to change with exercise alone, because
they are likely to be affected by psychosocial and environmen-
tal factors as well as by physical ability.
21
Although weight-bearing exercise is less commonly used in
current clinical practice than nonweight-bearing exercise, it
was not associated with adverse effects in this study. No falls
were reported while doing the exercises, and subjects did not
report more pain while doing the weight-bearing exercise pro-
gram than what was reported by subjects doing the non
weight-bearing program. Subjects in the exercise groups also
had a comparable level of compliance with the prescribed
exercise program, suggesting a similar acceptance of the exer-
cises. Subjects in the weight-bearing group found the exercises
moderately more difcult, which probably reects the use of
greater muscle forces and more muscle groups to gain and
maintain the upright position required for the weight-bearing
exercises. It is possible that the weight-bearing exercises had a
greater effect on the physical outcome measures purely as a
result of this greater level of difculty.
This exercise program was conducted in the home, with little
supervision from the physical therapists, so it was not possible
to gather reliable information about the actual intensity of
training, that is, the number of repetitions completed or the
duration of exercise sessions. As a result, we cannot provide
information about the amount of practice required to produce
the improvements in motor function that were found.
Table 2: Strength Between-Group Comparisons
Outcome by Group Preintervention 1 Month 4 Months
Hip abduction affected leg
WBE 47.829.8 55.728.1 57.927.9
NWBE 49.925.7 54.126.0 56.324.3
Control 46.126.0 52.425.4 53.028.0
Hip abduction nonaffected leg
WBE 63.128.5 66.625.0 72.231.3
NWBE 61.429.1 65.927.0 68.827.2
Control 60.831.3 62.930.9 64.830.1
Hip exion affected leg
WBE 39.823.6 47.625.3 50.428.3
NWBE 38.821.2 45.824.8 49.323.1
Control 38.222.1 41.522.4 43.326.4
Hip exion nonaffected leg
WBE 52.923.9 57.924.9 60.829.4
NWBE 56.424.1 60.425.5 62.925.3
Control 53.426.2 59.126.5 55.326.7
Knee extension affected leg
WBE 124.562.7 141.056.3 152.975.9*
NWBE 108.848.1 119.647.3 125.647.2
Control 106.453.8 113.461.0 112.973.0
Knee extension nonaffected leg
WBE 146.671.2 162.970.0 172.079.6*
NWBE 146.961.4 153.368.5 152.260.3
Control 139.864.5 145.580.2 139.278.9
NOTE. Values are mean SD (N). Subject numbers at 1 month: WBE, n33; NWBE, n37; control, n36; 4 months: WBE, n32; NWBE, n34;
control, n36. At 4 months, post hoc tests revealed differences between the control and WBE groups for knee extension strength for both the
affected leg (mean difference, 31.8N; 95% condence interval [CI], 9.554.0; P.02) and the nonaffected leg (mean difference, 27.2N; 95% CI,
5.648.7; P.04).
714 EXERCISE AFTER HIP FRACTURE, Sherrington
Arch Phys Med Rehabil Vol 85, May 2004
Nineteen percent of subjects in the exercise groups who
underwent the nal assessment were not doing the prescribed
exercises at that time. This compliance rate is consistent with
other exercise intervention studies
22
and is not surprising, be-
cause no contact was made with the subjects between the 1-
and 4-month assessments. A program that incorporated more
visits or follow-up phone calls might have produced a higher
compliance rate. A more intense program might also have led
to greater between-group differences in the outcome measures.
The WBE program was designed to reect the way muscles
work during daily weight-bearing tasks, such as standing up,
walking, reaching, and stair-climbing. Larger improvements
for the WBE group were found for the measures that reected
daily task requirements (functional reach, step test, sit-to-stand
ability, walking aid use, and step-up ability); there were no
measures in which the NWBE group improved more than the
weight-bearing group. It therefore appears that exercises that
more closely approximate daily functional tasks are a better
choice for this population.
The external validity of this study is likely to be high.
Subjects were representative of those with fractured hips, peo-
ple with a wide range of ages (5795y) were included, and
Table 3: Balance Between-Group Comparisons
Outcome by Group Preintervention 1 Month 4 Months
Step test affected leg (steps)
WBE 7.05.4 10.06.4* 11.06.3*
NWBE 7.77.1 8.37.3 9.46.7
Control 8.36.5 9.47.0 9.07.3
Step test nonaffected leg (steps)
WBE 7.85.2 11.46.0* 11.75.9*
NWBE 7.65.8 8.76.5 9.15.8
Control 8.96.3 10.36.4 9.97.2
Functional reach (cm)
WBE 17.56.8 21.09.0 24.88.8*
NWBE 18.49.1 18.77.4 19.98.1
Control 17.88.7 19.99.0 19.410.0
Sway distance oor (mm)
WBE 76.445.8 78.847.7 79.046.0
NWBE 75.953.6 81.353.1 70.052.8
Control 90.350.3 89.254.4 89.859.9
Sway distance foam (mm)
WBE 136.681.2 128.580.5 115.773.3
NWBE 146.3108.1 142.5111.2 122.084.1
Control 143.595.6 128.885.2 129.079.4
NOTE. Values are mean SD. Subject numbers at 1 month: WBE, n33; NWBE, n37; control, n36; 4 months: WBE, n33; NWBE, n35;
control, n36.
*P .05. Univariate tests of between-group differences on change scores. At 4 months, post hoc tests revealed differences between the WBE
and control groups for the step test on both the affected leg (mean difference, 2.8 steps; 95% CI, 0.94.8; P.01) and the nonaffected leg (mean
difference, 2.6 steps; 95% CI, 0.94.7; P.008) and for the functional reach (mean difference, 5.9cm; 95% CI, 2.19.6; P.004) and between the
WBE and NWBE groups for the step test on both the affected (mean difference, 2.9 steps; 95% CI, 0.94.8; P.01) and nonaffected legs (mean
difference, 3.1 steps; 95% CI, 1.34.8; P.003) and for the functional reach (mean difference, 7.1cm; 95% CI, 3.410.9; P.001).
Table 4: Gait Between-Group Comparisons
Outcome by Group Preintervention 1 Month 4 Months
Time to walk 6m at comfortable pace (s)
WBE 14.310.0 12.39.0 11.89.2
NWBE 15.913.2 14.211.4 13.212.9
Control 14.210.9 13.312.1 13.211.7
No. of steps taken in 6m at comfortable pace
WBE 19.310.1 18.710.8 18.311.0
NWBE 17.66.4 17.05.2 17.37.1
Control 17.18.8 17.711.1 17.29.4
Time to walk 6m at fast pace (s)
WBE 10.48.6 9.28.4 9.29.1
NWBE 12.310.8 11.010.0 9.910.8
Control 11.210.2 11.915.7 11.011.6
No. of steps taken in 6m fast pace
WBE 17.19.8 16.210.3 16.310.7
NWBE 16.15.8 15.65.3 15.76.0
Control 15.98.3 16.712.8 16.39.9
NOTE. Values are mean SD. Subject numbers at 1 month: WBE, n33; NWBE, n37; control, n37; 4 months: WBE, n33; NWBE, n36;
control, n36.
715 EXERCISE AFTER HIP FRACTURE, Sherrington
Arch Phys Med Rehabil Vol 85, May 2004
there were few exclusion criteria. In addition, subjects were
recruited from several hospitals, both public and private care,
and from several different settings (orthopedic wards, rehabil-
itation wards, physical therapy departments). The subjects had
received several different treatment approaches before entering
the study and represented a range of socioeconomic and ethnic
backgrounds.
CONCLUSIONS
This study indicates that a home-exercise program with
limited supervision can be of benet after hip fracture. Weight-
bearing exercise produced greater improvements than non
weight-bearing exercise or no exercise, particularly in balance
and functional performance. All participants had completed the
usual postfracture care, so it is evident that they had the
potential for further improvements in physical functioning,
particularly with a weight-bearing exercise program.
Acknowledgments: This study is part of the National Health and
Medical Research Councils Prevention of Injuries in Older People
Health Partnership research program. Staff at 6 Sydney hospitals
(Bankstown-Lidcombe, Liverpool, War Memorial, Balmain, Green-
wich, Hunters Hill Private) assisted with the recruitment of subjects.
Jenny Jacka assisted with nancial management. Pat Pamphlett and
Pernille Jensen assisted with exercise prescription and data collection.
Dr. Hylton Menz made useful comments on an earlier draft of this
manuscript.
References
1. Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly
persons living in the community. N Engl J Med 1988;319:1701-7.
2. Cummings SR, Black DM, Rubin SM. Lifetime risks of hip, Colles,
or vertebral fracture and coronary heart disease among white post-
menopausal women. Arch Intern Med 1989;149:2445-8.
3. Johnell O. The socioeconomic burden of fractures: today and in
the 21st century. Am J Med 1997;103(2A):20S-25S; discussion
25S-26S.
4. Marottoli RA, Berkman LF, Cooney LM Jr. Decline in physical
function following hip fracture. J Am Geriatr Soc 1992;40:861-6.
5. Sherrington C, Lord SR. Home exercise to improve strength and
walking velocity after hip fracture: a randomized controlled trial.
Arch Phys Med Rehabil 1997;78:208-12.
6. Tinetti ME, Baker DI, Gottschalk M, et al. Home-based multicom-
ponent rehabilitation program for older persons after hip fracture: a
randomized trial. Arch Phys Med Rehabil 1999;80:916-22.
7. Tinetti ME, Baker DI, Gottschalk M, et al. Systematic home-base
physical and functional therapy for older persons after hip frac-
ture. Arch Phys Med Rehabil 1997;78:1237-47.
8. Rush S. Rehabilitation following ORIF of the hip. Top Geriatr
Rehabil 1996;12:38-45.
9. Carr J, Shepherd R. Neurological rehabilitation: optimizing motor
performance. Oxford: Butterworth-Heinemann; 1998.
10. Jones D, Rutherford O, Parker D. Physiological changes in skel-
etal muscle as a result of strength training. Q J Exp Physiol
1989;74:233-56.
11. Lord SR, Clark RD, Webster IW. Physiological factors associated
with falls in an elderly population. J Am Geriatr Soc 1991;39:
1194-200.
12. Bohannon RW. Test-retest reliability of hand-held dynamometry
during a single session of strength assessment. Phys Ther 1986;
66:206-9.
13. Duncan PW, Weiner DK, Chandler J, Studenski S. Functional
reach: a new clinical measure of balance. J Gerontol A Biol Sci
Med Sci 1990;45:M192-7.
14. Hill KD, Bernhardt J, McGann AM, Maltese D, Berkovits D. A
new test of dynamic standing balance for stroke patients: reliabil-
ity, validity and comparison with healthy elderly. Physiother Can
1996;48:257-62.
15. Winograd CH, Lemsky CM, Nevitt MC, et al. Development of a
physical performance and mobility examination. J Am Geriatr Soc
1994;42:743-9.
16. Nugent JA, Schurr KA, Adams RD. A dose-response relationship
between amount of weight-bearing exercise and walking outcome
following cerebrovascular accident. Arch Phys Med Rehabil
1994;75:399-402.
17. Cifu DX. Rehabilitation of fractures of the hip. Phys Med Rehabil
State Art Rev 1995;9:125-39.
18. Hochberg Y. A sharper Bonferroni procedure for multiple tests of
signicance. Biometrika 1988;75:800-2.
19. Lord SR, Ward JA, Williams P, Anstey K. Physiological factors
associated with falls in older community-dwelling women. J Am
Geriatr Soc 1994;42:1110-7.
20. Lord SR, Ward JA, Williams P, Strudwick M. The effect of a 12-month
exercise trial on balance, strength, and falls in older women: a random-
ized controlled trial. J Am Geriatr Soc 1995;43:1198-206.
21. Keysor JJ, Jette AM. Have we oversold the benet of late-life
exercise? J Gerontol A Biol Sci Med Sci 2001;56:M412-23.
22. Dishman RK. Introduction: Consensus, problems, and prospects.
In: Dishman RK, editor. Advances in exercise adherence. Cham-
paign: Human Kinetics; 1994. p 1-27.
23. Sonn U, Asberg KH. Assessment of activities of daily living in the
elderly. A study of a population of 76-year-olds in Gothenburg,
Sweden. Scand J Rehabil Med 1991;23:193-202.
24. Holden MK, Gill KM, Magliozzi MR. Gait assessment for neu-
rologically impaired patients. Standards for outcome assessment.
Phys Ther 1986;66:1530-9.
Supplier
a. SPSS Inc, 233 S Wacker Dr, 11th Fl, Chicago, IL 60611.
Table 5: Functional Performance Between-Group Comparisons
Outcome by Group Preintervention 1 Month 4 Months
Timed sit-to-stand 5 (s)
WBE 25.418.5 20.014.1 18.012.4*
NWBE 28.120.4 24.920.3 20.312.7
Control 21.712.8 22.521.8 23.215.4
Timed supine-to-sit (s)
WBE 4.63.9 3.83.6 3.43.5
NWBE 6.88.7 5.45.5 5.59.2
Control 4.53.6 3.73.4 4.13.5
PPME total score
WBE 9.42.1 10.22.3 10.32.3
NWBE 9.52.0 9.91.8 10.51.5
Control 9.81.8 10.21.7 10.11.8
NOTE. Values are mean SD. Subject numbers at 1 month: WBE,
n33; NWBE, n37; control, n34; 4 months: WBE, n33; NWBE,
n36; control, n36.
*P.05. Univariate tests of between-group differences on change
scores. At 4 months, post hoc tests revealed differences for timed
sit-to-stand between the WBE and control groups (mean difference,
8.1s; 95% CI, 3.412.8; P.001) and between the NWBE and control
groups (mean difference, 5.8s; 95% CI, 1.210.5; P.02).
716 EXERCISE AFTER HIP FRACTURE, Sherrington
Arch Phys Med Rehabil Vol 85, May 2004