Bone 39 (2006) 1203 1209

www.elsevier.com/locate/bone

Site-specific response of bone to exercise in premenopausal women

Kerri M. Winters-Stone , Christine M. Snow
Bone Research Laboratory, Department of Nutrition and Exercise Science, Oregon State University, Corvallis, OR 97331, USA
Received 3 March 2006; revised 13 May 2006; accepted 16 June 2006
Available online 28 July 2006

Abstract
We studied the response of bone at specific skeletal sites to either lower body exercise alone or complemented with upper body exercise in
premenopausal women. Thirty-five exercisers and 24 age-matched controls completed the 12-month study. Exercising women (N = 35) were
randomly assigned to either lower body resistance plus jump exercise (LOWER) (N = 19) or to lower and upper body resistance plus jump exercise
(UPPER + LOWER) (N = 16). Exercisers trained three times per week completing 100 jumps and 100 repetitions of lower body resistance with or
without 100 repetitions of upper body resistance exercise at each session. Intensity for lower body exercise was increased using weighted vests for
jump and resistance exercises, respectively. Intensity for upper body exercise was increased using greater levels of tautness in elastic bands. Bone
mineral density (BMD) at the total hip, greater trochanter, femoral neck, lumbar spine and whole body were measured by dual energy X-ray
absorptiometry (Hologic QDR-1000/W) at baseline, 6 and 12 months. Data were analyzed first including all enrolled participants who completed
follow-up testing and secondly including only those women whose average attendance was 60% of prescribed sessions. Group differences in 12month %change scores for BMD variables were analyzed by univariate ANCOVA adjusted for baseline differences in age. Post hoc tests were
performed to determine which groups differed from one another. Initial analysis showed significant differences in greater trochanter BMD between
each exercise group and controls, but not between exercise groups (2.7% 2.5% and 2.2% 2.8% vs. 0.7% 1.7%, for LOWER and UPPER
+ LOWER vs. controls, respectively; p < 0.02) and near significant group differences at the spine (p = 0.06). Excluding exercisers with low
compliance, group differences at the greater trochanter remained, while spine BMD in UPPER + LOWER was significantly different from
LOWER and controls, who were not significantly different from one another (1.4% 3.9% vs. 0.9% 1.7% and 0.6% 1.8%, for UPPER
+ LOWER vs. LOWER and controls, respectively; p < 0.05). No significant differences among groups were found for femoral neck, total hip or
whole body BMD. Our data support the site-specific response of spine and hip bone density to upper and lower body exercise training,
respectively. These data could contribute to a site-specific exercise prescription for bone health.
2006 Elsevier Inc. All rights reserved.
Keywords: Bone mineral density; Osteoporosis; Premenopausal; Resistance; Impact; Site-specific

Introduction
Hip and spine fractures are two of the most commonly
fractured sites related to osteoporosis and bone mineral density
(BMD) measured at these sites is an index of bone health and
fracture risk. There are common determinants of axial bone
mineral density (BMD), such as age; however, the relative
influence of other factors may vary from site to site. For
example, the spine is more sensitive than the hip to changes in
circulating sex hormones because of the higher proportion of
Corresponding author. School of Nursing, Mailcode: SN-ORD, Oregon
Health and Science University, Portland, OR 97239, USA. Fax: +1 503 418
0903.
E-mail address: wintersk@ohsu.edu (K.M. Winters-Stone).
8756-3282/$ - see front matter 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.bone.2006.06.005

trabecular bone [1,2]. Physical activity is known to influence

bone as well though the relative effects of varying modes of
activity may be site specific.
Wolffe's law states that bone responds to the forces placed
upon it, such that BMD is greatest in regions of high force
application. Perhaps the most commonly recognized illustration
of Wolffe's law is the consistent difference between the playing
and non-playing arm of racquet sports athletes, where the
playing arm is not only denser, but structurally stronger than the
non-playing arm [3,4]. Other contralateral differences in BMD
related to the asymmetric loading patterns during activity have
been reported in several groups of athletes [57].
To date, most exercise interventions aimed at improving
bone health in women have been general rather than specific to
either the hip or spine. While it is useful to define a general

1204

K.M. Winters-Stone, C.M. Snow / Bone 39 (2006) 12031209

exercise prescription for overall bone health [8], a targeted

program may be more effective at improving site-specific
BMD. The limited exercise programs specific to the hip or spine
have shown that impact (jump) training increased hip, but not
spine BMD [912]; and that a 3-year back strength training
program, did not increase spine BMD [13]. To our knowledge,
no study has been designed to identify site-specific exercises for
each clinical site within the same study population.
We previously report that impact plus lower body resistance
training improved hip, but not spine, bone density in
premenopausal women [12]. To also study the response of the
spine to specific training, we added upper body exercise to the
lower body program in order to provide a targeted stimulus to
improve spine BMD. We hypothesized that women who
performed both upper and lower body exercise would increase
BMD at both the spine and the hip [1416] whereas women
who performed lower body exercise only would demonstrate an
increase in BMD at the hip but not the spine [911].

repetitions of lower body resistance exercises. Repetitions were performed in

succession with approximately 1530 s of rest between jumps and 23 min of
rest between sets of resistance exercise. Jumping routines varied in type and
height, but totaled 100 for each session. Jumps were performed in the following
ways: (1) off the ground, (2) off 8-in. wooden boxes, (3) in the forward and side
directions, and (4) in single- or double-leg stances. Participants performed the
jumps on 2-in. gymnastics mats and were instructed to jump with shoes off and
to land on both feet with approximately 30 of knee flexion. Lower body
exercises (squats, lunges, and calf raises) were performed immediately
following jumps. In general, each session consisted of 3 sets of 1012 squats,
6 sets of 1012 lunges (2 sets in two directions), and 2 sets of calf raises.
Intensity for both jump and resistance exercise was increased using weighted
vests and calculated as a percentage of body weight (%BW), such that each
woman exercised at the same relative intensity. Jump and resistance intensity
were progressively increased over the program to a final intensity of 10%BW for
jumps and 13%BW for resistance exercises. We have previously reported that
these lower body resistance exercises produce ground reaction forces of 45
times body weight measured on a force plate in the laboratory [12]. Since
exercises were performed on a gymnastics mat, actual GRFs during training may
have been attenuated by 0.5 to 0.75 body weights based on comparisons of
force production between landings on unforgiving versus on area-elastic
surfaces, such as gymnastics mats [17]. Women recorded their training on
individual logs kept in the exercise room.

Materials and methods

Participants
Women residing in Corvallis, Oregon, and surrounding areas were solicited
to participate in a 12-month exercise training program. To be eligible, women
had to be premenopausal at study enrollment (912 menstrual cycles in the
previous 12 months), and were excluded for any of the following: history of
chronic disease known to affect bone metabolism or exercise capacity, smoking,
breast feeding, intention to become pregnant within the next year, preexisting
musculoskeletal condition, or regular participation (2 times per week) in
resistance training or in activities including high-impact movements (e.g.,
volleyball, basketball). Women who currently participated in aerobic activities,
such as walking or jogging, were not excluded and were encouraged to maintain
this habitual activity throughout the study period. Sixty percent of interested
women did not meet eligibility criteria and thus, 42 women were initially
enrolled into the exercise program and began the 12-month study.

Study design
The initial recruitment effort aimed to enroll women into the exercise
program. We subsequently recruited a control group of women matched to the
exercise group on age and hip and spine BMD. The same eligibility criteria for
enrollment in the exercise program were applied to controls. The Oregon State
University Institutional Review Board approved the study and all participants
provided written informed consent. To evaluate the site-specific response of
bone to training, women within the exercise group were randomly assigned to
perform lower body resistance plus jump exercise (LOWER) or lower and upper
body resistance plus jump exercise (UPPER + LOWER).

Exercise training
Exercise participants were asked to attend three exercise sessions per week
with at least 1 day of rest between sessions. Classes were held on the Oregon
State University campus. Both exercise groups trained at the same time and in
the same setting. All participants performed the warm-up and jumping phases as
a large group, then broke into their respective exercise groups for the resistance
exercise phase, and rejoined for a large group cool-down. Two exercise trainers
led each exercise session so that a trainer was available for each group during the
resistance phase of training.
Lower body exercise
Women in both training groups performed lower body exercise. The lower
body training program consisted of 9 sets of 1012 jumps and 9 sets of 1012

Upper body exercise

Following jumps, participants randomized to the UPPER + LOWER group
performed 3 sets of 812 repetitions of upper body exercises, with
approximately 12 min of rest between sets. Exercises consisted of the
following: upright row, one-arm row, latissimus dorsi pull-down, chest press,
chest fly, biceps curl and triceps extension. In order to be time efficient and to
have both groups complete resistance exercises at the same time, upper and
lower body exercises were often combined into one movement. Examples of
combined exercise include the following: biceps curl or upright row with squat,
tricep extension or one-arm row with side lunge, lat pull-down or chest press/fly
with front lunge. One type of exercise for each major muscle group was
performed at each session, though over the week specific exercises were varied.
Resistance to the exercises was delivered via elastic bands (TheraBand, Inc.)
selected to achieve an intensity of 812 Repetition Maximum (RM), within
the recommended range of repetitions for improving muscle fitness in novice
resistance exercisers [18]. When participants could perform more than 12
repetitions with a given band, the band resistance was increased so that women
could only perform 8 repetitions. This volume of elastic band training has been
reported to produce significant increases in strength in premenopausal women
[19].

Height and weight

Measurement of height and weight was conducted with participants in
regular dress clothing, but without shoes. Standing height was measured to the
nearest 0.5 cm using a wall-mounted stadiometer. Body weight was measured on
a digital scale to the nearest 0.1 kg.

Bone mineral density and body composition

Bone mineral density (BMD: g/cm2) of the total hip, greater trochanter,
femoral neck, lumbar spine (L24), and whole body was measured via DXA
(Hologic QDR 1000-W, software version 4.74) at 0, 6 and 12 months. Lean and
fat masses of the whole body, arms and legs were determined from whole body
scans. The same individual conducted all scans and analyses. In-house
coefficients of variation (CV) on a subsample of women (N = 10) similar to
our study population are <1.0% for hip and spine measures and <1.5% for body
composition measures [12].

Nutrient intake
Nutrient intake was assessed in order to control for the potential confounding
effects of baseline calcium intake or changes in calcium intake on BMD
outcomes. Macro- and micronutrient intake was assessed from 3-day self-

K.M. Winters-Stone, C.M. Snow / Bone 39 (2006) 12031209

reported food intake records at baseline and 12 months. Participants were
instructed to record the type and amount of all food and drink consumed over 3
consecutive days, including 2 weekdays and 1 weekend day. Participants were
also instructed to include nutritional supplements ingested over these 3 days and
these were included in the estimate of nutrient intake. Daily consumption of total
energy (kilocalories), carbohydrate (g), protein (g), fat (g), and calcium (mg)
was estimated using the Food Processor II nutrient analysis software program
(Salem, OR).

Statistical analysis
Descriptive data are presented as mean standard deviation; graphical data
are presented as mean standard error. Initial differences between the three
study groups were determined using one-way ANOVA on baseline data. Oneway ANCOVA was also used to evaluate group differences in 12-month percent
change in BMD at each bone site and for body composition, adjusting for any
significant baseline group differences in participant characteristics (age, height,
weight, calcium intake or baseline BMD). Follow-up post hoc analyses on
significant ANCOVAs were performed using Least Significant Difference
(LSD) method to determine which group contrasts were significantly different.
Significance levels were set at p < 0.05.
Data were initially analyzed for all participants who enrolled in the project
and completed follow-up testing, including those who had low compliance to
exercise training. To better determine the effect of habitual exercise training on
dependent measures, we subsequently re-ran analyses excluding those women
who did not attend class regularly. We defined low compliance by identifying
the natural breakpoint in percent of class attendance among all participants.
There was a clear break in % attendance at a value of 60%, where 80% of our
participants had attendance above 60% and 20% (N = 7; 5 participants from the
LOWER and 2 participants from the UPPER + LOWER group) had attendance
below this mark. In this analysis, group sizes for each exercise group with
compliers only, were reduced to N = 14. Repeating analyses on a restricted
sample based on compliance has been done in other studies of exercise training
and bone [20,21].
To evaluate the time course of the bone response to training, we also
calculated % change in BMD for 06 months, and 612 months (% change data
for these time points not presented. Absolute BMD at 6 months is included in
Table 2). We then repeated the analysis of variance to determine whether group
differences could be detected over these shorter time frames. All analyses were
performed using the SPSS statistical software program, version 11.0.

Results
Participants
Thirty-five exercisers (LOWER, n = 19; UPPER + LOWER,
n = 16) and 24 controls were tested after 12 months. Seven
women in the exercise group (LOWER, n = 2; UPPER
+ LOWER, n = 5) withdrew within the first month of the exercise
program for the following reasons: increased musculoskeletal
pain (n = 3), relocation (n = 1); pregnancy (n = 1); disinterest
(n = 2). Data for these women were not included in statistical
analyses. All controls completed both baseline and 12-month
visits. Baseline data on dependent measures between those who
completed both testing periods of the study and those who
withdrew were not significantly different.
Participants were mature premenopausal women of average
height, weight, body composition and bone health (Table 1).
Daily intake of calcium was adequate in all groups based on
baseline values (Table 1) and did not differ among groups at
baseline or over time. Age differed significantly between groups
and was included as a covariate in statistical analyses. Groups
were similar on remaining variables. Age differences disap-

1205

Table 1
Baseline descriptive characteristics of participants completing both baseline and
follow-up testing (N = 59)

Age (years)
Height (m)
Weight (kg)
Body fat (%)
Calcium intake (mg/d)
Total hip Z-score
Lumbar spine Z-score
a

LOWER
(N = 19)
Mean SD

UPPER + LOWER
(N = 16)
Mean SD

CONTROL
(N = 24)
Mean SD

38.3 3.8 a
1.64 0.01
65.0 12.5
27.6 6.0
791 280
0.06 0.92
0.28 1.01

41.3 3.8
1.66 .06
68.5 8.3
28.1 4.4
906 313
0.04 0.82
0.52 1.06

40.5 3.5
1.67 .06
66.9 13.1
26.9 8.4
1060 337
0.18 0.67
0.71 1.16

Significant group differences, LOWER vs. UPPER + LOWER; p < 0.05

peared when low compliers were excluded and was not adjusted
for in these sample-restricted analyses. Overall compliance to
exercise among all completers averaged 71% and did not differ
significantly between groups (74% vs. 69% for UPPER +
LOWER and LOWER, respectively). Compliance increased to
77% for each group when data from low compliers were
excluded.
Though we did not include measures of upper and lower
body strength to assess the effectiveness of training, previous
studies have shown that this volume of elastic resistance and
weighted vest training produce significant increases in upper
and lower body strength, respectively [12,19]. The progression
in training intensity over the 12-month program serves as an
additional indicator that women improved muscle fitness. The
sets and repetitions of the training program were near constant
over the course of training and consistent with recommended
ranges for strength training. Intensity of exercises was
progressively increased from 0% to 10%13% of body weight
for lower body exercises over 12 months. For upper body
training, women increased intensity by using thicker elastic
bands, moving from yellow to red to green bands over the
course of training. Using force estimates for these color bands,
at a given % elongation, the force production increases by
50%60% with each increase in band color [22].
Training effects on body composition
Both exercise groups increased total body lean mass and
decreased total body fat mass over 12 months; whereas control
participants lost lean mass and gained fat mass (Table 3).
Significant differences among groups were only detected for
12-month changes in fat mass and post hoc tests revealed
significant differences between the UPPER + LOWER group
and controls. Differences remained when analyses were
repeated excluding participants with low compliance, though
fat mass differences between LOWER and controls now
approached significance (p = 0.06).
We also derived arm and leg body composition from the
regional body composition analysis (Table 3). Lean mass of the
arms increased slightly among all groups, while fat mass of the
arms slightly decreased in UPPER + LOWER and increased in
LOWER and controls. Leg lean mass increased more in both
exercise groups than in controls, while leg fat mass decreased in

1206

K.M. Winters-Stone, C.M. Snow / Bone 39 (2006) 12031209

exercisers and increased in controls. None of the regional body

composition changes differed significantly among groups in
analyses with all participants or in those with compliant
exercisers only.
Training effects on BMD
After adjusting for baseline differences in age, both exercise
groups increased BMD at the greater trochanter (Table 2)
significantly more than controls. The UPPER + LOWER group
increased spine BMD compared to decreases in LOWER and
control groups. Group differences were significant for % change
in BMD of the greater trochanter (LOWER: +2.6% 2.5%,
UPPER + LOWER: + 2.2% 2.8%, CON: + 0.7% 1.7%;
p < 0.02) and approached significance at the lumbar spine
(LOWER: 0.3% 1.7%, UPPER + LOWER: +1.3% 3.7%,
CON: 0.5% 1.9%; p = 0.06). In post hoc analyses trochanter
BMD changes were greater in exercise groups than controls
(p < 0.01). At the lumbar spine, BMD was significantly different
between UPPER + LOWER vs. CON (p = 0.03), nearly significantly different between UPPER + LOWER and LOWER
(p = 0.07), and not different between LOWER vs. CON
(p = 0.9).
Analyses excluding low compliers were not adjusted for
baseline age because is in this restricted analysis age differences
were no longer different among groups. Significant group
differences at the greater trochanter remained, and differences in
Table 2
Baseline, 6-month (mid-training) and 12-month (post-training) BMD values and
12-month percent change (% ) for participants completing baseline and
follow-up testing (N = 59)

Total hip
Baseline
6 months
12 months
% (012)
Femoral neck
Baseline
6 months
12 months
% (012)
Greater trochanter
Baseline
6 months
12 months
% (012)
Lumbar spine
Baseline
6 months
12 months
% (012)
Whole body
Baseline
6 months
12 months
% (012)

LOWER N = 19 UPPER + LOWER

Mean SD
N = 16 Mean SD

CONTROL N = 24
Mean SD

0.917 0.114
0.924 0.115
0.925 0.110
1.1 1.6

0.921 0.107
0.931 0.109
0.935 0.104
1.5 1.5

0.940 0.082
0.943 0.080
0.946 0.080
0.7 1.6

0.835 0.118
0.838 0.118
0.843 0.118
1.0 3.3

0.830 0.098
0.829 0.104
0.838 0.100
1.0 3.2

0.846 0.091
0.845 0.088
0.847 0.096
0.0 2.2

0.690 0.096
0.701 0.096
0.707 0.095
2.6 2.5 a

0.706 0.096
0.715 0.093
0.720 0.091
2.2 2.8 a

0.718 0.064
0.721 0.064
0.723 0.063
0.7 1.7

1.088 0.107
1.081 0.110
1.085 0.108
0.3 1.9

1.093 0.125
1.098 0.117
1.105 0.118
1.3 3.7 a

1.126 0.127
1.121 0.124
1.119 0.121
0.5 1.8

1.091 0.081
1.096 0.080
1.097 0.077
0.7 1.3

1.115 0.068
1.123 0.072
1.124 0.067
0.9 1.3

1.103 0.051
1.103 0.053
1.116 0.055
0.0 1.4

Fig. 1. 12-month percent change in BMD among compliant exercisers and

controls. Data are presented as mean standard error. Abbreviations: Fneck =
femoral neck; Troch = greater trochanter; Spine = lumbar spine (L2L4).
*Significantly different from controls, p < 0.05. Significantly different between
LOWER and UPPER + LOWER, p < 0.05.

spine BMD were significant (Fig. 1). In post hoc analyses at the
spine, UPPER + LOWER gained BMD at this site compared to a
decrease in BMD in LOWER (p < 0.05).
Analyses on all participants and then repeated on compliers
only failed to detect significant differences in BMD among
groups at the total hip, femoral neck or whole body.
Time course of bone changes
Inclusion of a mid-training BMD evaluation allowed us to
evaluate the time course of the BMD response to training (Table
2). Increases in BMD at the greater trochanter, total hip and
whole body were greater in the first half of training and
continued through the second half, though at a slower rate.
Femoral neck increased most in the second half of training. Spine
BMD increased more in the latter half of training compared to the
first 6 months in UPPER + LOWER. Analyses of covariance on
% change in BMD in each the first and second halves of training
detected group differences in whole body BMD within the first
6 months of training that did not persist at 12 months.
Discussion

a
Significant group differences from control, using 12 month % change
adjusted for age, p < 0.05.

Our study demonstrated that bone responds to site-specific

exercise. That is, women who added upper body resistance
exercise to a routine of lower body resistance plus jump training
increased BMD of both the hip and spine. In contrast, women
who performed lower body training only, increased hip but not
spine BMD. To our knowledge, we are the first to compare
exercise that targeted the hip only to exercise that targeted both
the hip and spine, sites of great clinical importance, in the same
cohort of premenopausal women. Our data also suggest that
compliance to training influences the extent of bone changes at
the hip and spine. Changes in hip BMD and differences in spine
BMD between groups were more pronounced when analyses
were limited to women who exercised regularly ( 60%
attendance at prescribed sessions), despite the lower sample
size (Fig. 1). Others also report greater BMD responses to
exercise when data from less or non-compliant exercisers are
excluded [20,21].

K.M. Winters-Stone, C.M. Snow / Bone 39 (2006) 12031209

BMD increased progressively over time in exercisers (Table

3), suggesting that BMD could continue to increase beyond
12 months with continued training, but must be confirmed in
future studies of longer duration. Our exercise program also
decreased fat mass in women who performed combined upper
and lower body exercise. Since the upper body exercises were
added to the lower body program in UPPER + LOWER, women
in this group performed more exercise and thus expended more
energy than LOWER. Lean mass and regional body composition improved in the exercise groups, but differences among
groups were not significant. Our control group gained both lean
and fat mass over the 12-month study, potentially obscuring our
ability to attribute the greater increases in lean body mass
among exercisers to our training program. The failure of our
program to affect either total or regional lean mass may not
necessarily indicate a lack of training effect on muscular fitness,
since muscle fitness increases can occur in the absence of
increases in lean mass due to neural adaptations [23].
Our study also had limitations. We compared a program of
lower body exercise to one of lower plus upper body exercise to
study site-specific responses of the hip and spine; however, we
did not include a group that performed upper body exercise
only. If the upper body exercises were adding any stimulus to
the hip, we would have expected dissimilar BMD changes at the
Table 3
Baseline (pre), 12-month (post) lean and fat mass (kg) and absolute change ()
for arm, leg and total body for participants completing baseline and follow-up
testing (N = 59) a
LOWER
N = 19
Mean SD
Arm lean mass
Pre
Post

Arm fat mass

Pre
Post

Leg lean mass

Pre
Post

Leg fat mass

Pre
Post

Total body lean

mass
Pre
Post

Total body fat mass

Pre
Post

UPPER + LOWER
N = 16 Mean SD

CONTROL
N = 24 a
Mean SD

3.7 0.7
3.9 0.6
0.2 0.2

4.1 0.6
4.2 0.6
0.1 0.2

4.0 0.7
4.1 0.6
0.2 0.2

2.1 0.7
2.2 0.9
0.1 0.5

2.3 0.6
2.2 0.5
.1 0.4

2.2 0.8
2.3 0.2
0.1 0.2

13.3 2.1
13.8 2.1
0.6 0.4

14.4 2.1
15.3 2.1
0.8 0.7

14.0 1.6
14.3 1.7
0.4 0.5

8.3 2.5
8.1 2.5
0.2 1.1

8.7 2.5
8.5 2.4
0.3 1.4

8.3 3.9
8.4 4.0
0.1 0.6

44.2 5.7
45.0 5.5
1.2 0.7

46.6 5.1
47.5 5.5
0.8 1.5

45.5 4.6
46.0 4.8
0.5 1.2

18.5 7.3
18.1 7.1
0.4 3.3

19.4 4.9
17.9 4.5
1.5 2.4 b

18.9 7.1
19.7 1.5
0.8 2.4

a
n = 22 for control group for regional body composition analyses due to
invalid regional body composition data.
b
Significant group differences from control based on age-adjusted ANCOVA,
p < 0.05.

1207

hip between our upper + lower and lower groups, yet both
groups increased similarly at each hip site and increases were
not significantly different (Table 2, Fig. 1). Lower body training
did not appear to influence the spine as the lower body only
training group (LOWER) had spine BMD losses similar to
controls. The other limitation of our study was the nonrandomized design. Despite our expectation that self-selection
would encourage greater adherence to an intense intervention,
our retention and compliance rates were on par with those
reported for randomized controlled trials [24]. In our study,
women who self-selected into exercise groups were no more
motivated to comply with the program compared to women who
volunteer for a randomized trial. To compensate for the quasiexperimental design, we adjusted for initial differences in
factors that could alter the bone response to training, such as age
or calcium intake, when baseline values differed significantly
between groups. We did not systematically track habitual
physical activity over the course of the study, thus it is possible
that our groups altered their activity levels over time. The lack
of change in BMD among controls suggests that this group did
not alter bone-loading activity over time (Table 2). We assume it
was unlikely that our exercisers increased their activity levels
above the training program, though we cannot rule out the
possibility that increased activity outside the intervention
contributed to BMD gains.
Much of what is known about site specificity of bone to
exercise is drawn from cross-sectional comparison or longitudinal observation of athletes that demonstrate greater BMD
in dominant or playing limbs compared to non-dominant or
non-playing limbs [3,4,25]. Site specificity of bone to loading
has also been examined in training studies by comparing
unilateral loading on one limb to the other, non-loaded limb that
serves as a control. Similar to limb-to-limb comparisons of
athletes who experience asymmetrical loading, these designs
also control for inter-individual differences. Three studies in
premenopausal women examined the effects of unilateral upper
[26] or lower extremity [25,27,28], training on limb-specific
BMD; however, only one study documented a training effect.
One unilateral training study in postmenopausal women
reported a significant effect of high intensity forearm and
lower body resistance exercise on BMD of the exercised
forearm and hip and no effect on the unloaded contralateral
limbs [29]. While these unilateral training studies provide
decent evidence that the bone response to exercise occurs where
the stresses are greatest, they do not provide evidence for the
difference in bone responses to site-specific exercise at the two
bone sites of most clinical relevance. Sinaki attempted to apply
discrete loading to the spine in a 3-year program of back
extension exercises, yet failed to find a significant training
effect on BMD. These interventions that target a single skeletal
site cannot identify site-specific exercises for the two clinically
relevant skeletal regions.
Our results are consistent with other exercise interventions
targeting BMD in premenopausal women. Resistance and
aerobic training programs increase spine BMD by an average
of 1.2%1.5% for each mode, respectively [30]. In our
program, women assigned upper body training performed

1208

K.M. Winters-Stone, C.M. Snow / Bone 39 (2006) 12031209

exercises using resistance tubing selected to provide a

moderatevigorous training intensity (812 RM). Intervention studies of lower body training only, such as jumps,
where ground impacts and perhaps eccentric muscle actions
at landing provide the largest bone stimulus, report BMD
increases of up to 4% at the hip, but no effect at the spine
[9,10]. The LOWER group in our study that performed
lower body training only increased femoral BMD at all sites,
ranging from 1%2.6% at sites within the hip (Table 2).
When impact training is done in conjunction with other
types of training that includes upper body movements, or
with whole body resistance training, spine responses are
observed [14,16,31], as were in our UPPER + LOWER group
(Table 2; Fig. 1).
A limitation across previous interventions, however, is that
the exercises that produced a specific stimulus at each skeletal
site could not be discerned. Effective resistance and/or aerobic
training programs that improved hip and/or spine BMD,
employed whole body training of the trunk and upper and
lower extremities. Intervention studies where loads are applied
to both skeletal regions can only speculate about which
exercises positively affected or failed to affect specific skeletal
sites. One can infer site specificity by comparing findings
across studies, but none of these studies precisely answers the
question about which exercises best affect specific skeletal
regions.
An optimal intervention for low bone mass at a particular
skeletal site should select exercises that have been identified
as improving bone density of that particular region. For
example, a woman identified with below normal bone mass
of the hip, but not the spine, could increase hip BMD from
as few as 10 jumps, 3 times per week [11]. Jumps would not
be effective for someone with below normal bone mass
isolated to the spine, whereas a regular resistance program of
35 key upper body exercises would best target this clinical
site. Prescribing a targeted and efficient program for the
individual raises the likelihood that each person benefits the
most from their time and effort dedicated to physical
activity.
Acknowledgments
This project was supported by funding from the National
Aeronautics and Space Association Graduate Research Program
and Life Fitness, Inc.
References
[1] Duan Y, Tabensky A, DeLuca V, Seeman E. The benefit of hormone replacement therapy on bone mass is greater at the vertebral
body than posterior processes or proximal femur. Bone 1997;21:
44751.
[2] Margerison C, Paton LM, Nowson CA, Ferdous HS, Kaymakci B, Wark
JD. Hormone replacement therapy and bone mineral density: a co-twin
approach. Menopause 2002;9:43642.
[3] Haapasalo H, Kannus P, Sievanen H, Heinonen A, Oja P, Vuori I. Longterm unilateral loading and bone mineral density and content in female
squash players. Calcif Tissue Int 1994;54:24955.

[4] Haapasalo H, Kannus P, Sievanen H, Pasanen M, Uusi-Rasi K, Heinonen

A, et al. Effect of long-term unilateral activity on bone mineral density of
female junior tennis players. J Bone Miner Res 1998;13:3109.
[5] Lee EJ, Long KA, Risser WL, Poindexter HB, Gibbons WE, Goldzieher J.
Variations in bone status of contralateral and regional sites in young
athletic women. Med Sci Sports Exerc 1995;27:135461.
[6] Alfredson H, Nordstrom P, Pietila T, Lorentzon R. Long-term loading and
regional bone mass of the arm in female volleyball players. Calcif Tissue
Int 1998;62:3038.
[7] McClanahan BS, Harmon-Clayton K, Ward KD, Klesges RC, Vukadinovich CM, Cantler ED. Side-to-side comparisons of bone mineral density in
upper and lower limbs of collegiate athletes. J Strength Cond Res
2002;16:58690.
[8] Kohrt WM, Bloomfield SA, Little KD, Nelson ME, Yingling VR.
American College of Sports Medicine Position Stand: physical activity and
bone health. Med Sci Sports Exercise 2004;36:198596.
[9] Bassey E, Ramsdale S. Increase in femoral bone mineral density in young
women following high impact exercise. Osteoporos Int 1994;4:725.
[10] Bassey E, Rothwell M, Littlewood J, Pye D. Pre- and postmenopausal
women have different bone mineral density responses to the same highimpact exercise. J Bone Miner Res 1998;13:180513.
[11] Kato T, Terashima T, Yamashita T, Hatnaka Y, Honda A, Umemura Y.
Effect of low-repitition jump training on bone mineral density in young
women. J Appl Physiol 2006;100(3):83943.
[12] Winters KM, Snow CM. Detraining reverses positive effects of exercise on
the musculoskeletal system in premenopausal women. J Bone Miner Res
2000;15:2495503.
[13] Sinaki M, Wahner HW, Bergstralh EJ, Hodgson SF, Offord KP, Squires
RW, et al. Three-year controlled, randomized trial of the effect of dosespecified loading and strengthening exercises on bone mineral density
of spine and femur in nonathletic, physically active women. Bone
1996;19:23344.
[14] Heinonen A, Kannus P, Sievanen H, Oja P, Pasanen M, Rinne M, et al.
Randomised controlled trial of effect of high-impact exercise on selected
risk factors for osteoporotic fractures. Lancet 1996;348:13437.
[15] Lohman T, Going S, Pamenter R, Hall M, Boyden T, Houtkooper L, et al.
Effects of resistance training on regional and total bone mineral density in
premenopausal women: a randomized prospective study. J Bone Miner Res
1995;10:101524.
[16] Friedlander AL, Genant HK, Sadowsky S, Byl NN, Gluer CC. A two-year
program of aerobics and weight training enhances bone mineral density of
young women. J Bone Miner Res 1995;10:57485.
[17] Fritz M, Peikenkamp K. Simulation of the influence of sports surfaces on
vertical ground reaction forces during landing. Med Biol Eng Comput
2003;41:117.
[18] Kraemer WJ, Adams K, Cafarelli E, Dudley GA, Dooly C, Feigenbaum
MS, et al. American College of Sports Medicine position stand.
Progression models in resistance training for healthy adults. Med Sci
Sports Exercise 2002;34:36480.
[19] Kraemer WJ, Keuning M, Ratamess NA, Volek JS, McCormick M, Bush
JA, et al. Resistance training combined with bench-step aerobics enhances
women's health profile. Med Sci Sports Exercise 2001;33:25969.
[20] Uusi-Rasi K, Kannus P, Cheng S, Sievanen H, Pasanen M, Heinonen A, et
al. Effect of alendronate and exercise on bone and physical performance
of postmenopausal women: a randomized controlled trial. Bone 2003;
33:13243.
[21] Kemmler W, Lauber D, Weineck J, Hensen J, Kalender W, Engelke K.
Benefits of 2 years of intense exercise on bone density, physical fitness,
and blood lipids in early postmenopausal osteopenic women: results of the
Erlangen Fitness Osteoporosis Prevention Study (EFOPS). Arch Intern
Med 2004;164:108491.
[22] Page P, Labbe A, Topp R. Clinical force production at Thera-Band elastic
bands. J Orthop Sports Phys Ther 2000;30:A478.
[23] McBride JM, Blaak JB, Triplett-McBride T. Effect of resistance exercise
volume and complexity on EMG, strength, and regional body composition.
Eur J Appl Physiol 2003;90:62632.
[24] Wolff I, van Croonenborg J, Kemper H, Kostense P, Twisk J. The effect of
exercise training programs on bone mass: a meta-analysis of published

K.M. Winters-Stone, C.M. Snow / Bone 39 (2006) 12031209

controlled trials in pre- and postmenopausal women. Osteoporos Int
1999;9:112.
[25] Vuori I, Heinonen A, Sievanen H, Kannus P, Pasanen M, Oja P. Effects of
unilateral strength training and detraining on bone mineral density and
content in young women: a study of mechanical loading and deloading on
human bones. Calcif Tissue Int 1994;55:5967.
[26] Heinonen A, Sievanen H, Kannus P, Oja P, Vuori I. Effects of unilateral
strength training and detraining on bone mineral mass and estimated
mechanical characteristics of the upper limb bones in young women.
J Bone Miner Res 1996;11:490501.
[27] Hawkins SA, Schroeder ET, Wiswell RA, Jaque SV, Marcell TJ, Costa K.
Eccentric muscle action increases site-specific osteogenic response. Med
Sci Sports Exercise 1999;31:128792.

1209

[28] Kannus P, Haapasalo H, Sievanen H, Oja P, Vuori I. The site-specific

effects of long-term unilateral activity on bone mineral density and content.
Bone 1994;15:27984.
[29] Kerr D, Morton A, Dick I, Prince R. Exercise effects on bone mass in
postmenopausal women are site-specific and load-dependent. J Bone
Miner Res 1996;11:21825.
[30] Wallace BA, Cumming RG. Systematic review of randomized trials of the
effect of exercise on bone mass in pre- and postmenopausal women. Calcif
Tissue Int 2000;67:108.
[31] Vainionpaa A, Korpelainen R, Leppaluoto J, Jamsa T. Effects of
high-impact exercise on bone mineral density: a randomized controlled trial in premenopausal women. Osteoporos Int 2005;16:
1917.