Detraining Reverses Positive Effects of Exercise on the Musculoskeletal System in Premenopausal Women


  • Kerri M. Winters,

    Corresponding author
    1. Bone Research Laboratory, Department of Exercise and Sport Science, Oregon State University, Corvallis, Oregon, U.S.A.
    • Address reprint requests to: Kerri Winters, Ph.D., HPES Department, P.O. Box 15095, Northern Arizona University, Flagstaff, AZ 86011, U.S.A.
    Search for more papers by this author
  • Christine M. Snow

    1. Bone Research Laboratory, Department of Exercise and Sport Science, Oregon State University, Corvallis, Oregon, U.S.A.
    Search for more papers by this author


We studied the effects of a 6-month withdrawal of exercise after 12 months of progressive impact (jump) plus lower body resistance training on risk factors for hip fracture in premenopausal women (age, 30-45 years). Twenty-nine women completed the 12-month training and detraining programs and were compared with 22 matched controls. Bone mineral density (BMD) at the greater trochanter, femoral neck, lumbar spine, and whole body and body composition (% body fat) were measured by dual energy X-ray absorptiometry (DXA; Hologic QDR-1000/W). Knee extensor and hip abductor strength were assessed via isokinetic dynamometry (Kin-Com 500H); maximum leg power was tested using a Wingate Anaerobic Power test; and dynamic postural stability was measured on a stabilimeter (Biodex). All measurements were conducted at baseline, 12 months and 18 months with an additional midtraining measurement of BMD. Exercisers trained three times per week in a program of 100 jumps and 100 repetitions of resistance exercises at each session. Intensity was increased using weighted vests to final values of 10% and 13% of body weight (BW) for jump and resistance exercises, respectively. Differences between groups from training were analyzed by repeated measures analysis of covariance (ANCOVA), adjusted for baseline values. Detraining effects were analyzed by comparing the changes from training with the changes from detraining using repeated measures analysis of variance (ANOVA). Baseline values were not significantly different between exercisers and controls. Percent change over the training period was significantly greater in the exercise group than in the control group at the greater trochanter (2.7 ± 2.5% vs. 0.8 ± 0.8%, respectively; p < 0.01) and approached significance at the femoral neck (1.2 ± 3.2% vs. −0.3 ± 1.9%, respectively; p = 0.06). Significant improvements also were observed in exercisers versus controls for strength and power with exercisers increasing 13-15% above controls, whereas stability was not different between groups. After 6 months of detraining, BMD and muscle strength and power decreased significantly toward baseline values, whereas control values did not change. We conclude that the positive benefits of impact plus resistance training on the musculoskeletal system in premenopausal women reverse when training is withdrawn. Therefore, continued training, perhaps at a reduced frequency and intensity, is required to maintain the musculoskeletal benefit from exercise that may lower fracture risk in later life.


THE APPROXIMATE 1.5 million osteoporosis-related fractures that occur each year in the United States carry a significant health and economic burden.(1) Hip fractures result in the most severe consequences with respect to morbidity, mortality, and financial cost.(2) Low bone mineral density (BMD) at the hip increases the risk of fracture, and it is estimated that each 1 SD decrease in BMD increases fracture risk 10%.(3) However, the majority of reported hip fractures are not caused by low BMD alone, but rather result from injury associated with a fall.(4) Thus, the combination of low BMD and a propensity to fall significantly increases the fracture risk profile of an individual. Poor lower extremity strength and power and instability are independently associated with increased fall risk.(5,6) As individuals age, declines in BMD, muscle mass, and physical function are observed. However, age-associated declines may be attributed partly to accompanying reductions in habitual physical activity, as inactivity, immobilization, and bed rest also lead to significant musculoskeletal and functional decrements. There is limited evidence that engaging in activities that apply high loads to the musculoskeletal system may reverse or slow these physiological and functional declines.(7,8)

Activities that impart moderate to high loads and high loading rates have been shown to build bone and muscle mass and strength in pre- and postmenopausal women.(7, 9–12) In premenopausal women, activities with an impact component, that is, high loading rate, such as jumping alone or in combination with aerobic step exercise, increased hip BMD, vertical jump ability, and cardiovascular fitness, (9–11) while high-intensity resistance training improved both hip and spine BMD and muscle mass and strength.(12,13) However, no study has investigated the effect of combined impact (jump) and resistance exercise on multiple risk factors for hip fractures (hip BMD, lower extremity strength and power, and stability) in mature premenopausal women. Moreover, few studies incorporate the principles of specificity, overload, and reversibility of exercise training in intervention trials. Yet, demonstration of the principles of training in intervention work increases the probability of a favorable outcome and the credibility of study findings. Two reports in young, premenopausal women examined the effects of unilateral upper or lower extremity training and detraining on limb-specific strength and BMD(14,15); however, neither indicated a training effect. The only reported exercise training and detraining study in postmenopausal women showed gains in bone mineral content (BMC) from training and subsequent reduction in BMC after detraining but did not examine other risk factors for fractures.(16)

Thus, in our 12-month impact (jump) plus resistance training and subsequent 6-month detraining program we asked the following research questions: (1) will 1 year of high impact and resistance training result in significant improvements in hip BMD, lean mass, strength, power, and stability in mature premenopausal women? and (2) will gains in BMD, lean mass, strength, power, and stability reverse when the exercise stimulus is withdrawn?



Premenopausal women between 30 and 45 years of age were solicited to volunteer in the 18-month exercise training and detraining program. Initially, 110 women responded to local newspaper and electronic mail announcements. Of the original 110 respondents, 62 were excluded for one or more of the following reasons: history of chronic disease known to affect bone metabolism or exercise capacity, percent body fat >40%, smoking, breast feeding, intention to become pregnant within the next year, irregular menstrual cycles, or regular participation in high-intensity resistance training or in activities including high-impact movements (e.g., volleyball and basketball). After initial screening, 46 women were eligible; however, at training onset, 5 women decided not to participate in the training program and 41 women began the 18-month study. Reasons for not beginning the training program were as follows: relocation (n = 1), disinterest in study (n = 1), diagnosis of fibromyalgia (n = 1), or physician concern over preexisting orthopedic problems (n = 2).

The original design of the study was to have each woman act as her own control by including a 12-month observation period to precede the training program. We felt this type of design would control for genetic variability among women better than a randomized trial. However, time and budget constraints forced us to abandon the original design and thus we recruited an additional control group of women who were matched to the exercise group on age, body weight (BW), and BMD. Controls were recruited in a similar manner as the exercise group and resulted in 45 initial respondents. The aforementioned exclusion criteria and matching restrictions were used to screen women for entry into the control group and 24 were recruited. The Oregon State University Institutional Review Board approved the study and all subjects provided written informed consent.

The training and detraining programs

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. If participants could not attend all three classes on campus because of travel or other constraints, they were given an at-home training program to complete for missed sessions. The training program consisted of 9 sets of 10-12 jumps and 9 sets of 10-12 repetitions of lower body resistance exercises. Repetitions were performed in succession, with approximately 15-30 s of rest between jump sets and 2-3 minutes of rest between resistance exercise sets. Proper form was encouraged when executing all exercises and thus, speed was de-emphasized in favor of safe and controlled movements.

A variety of jumping routines was used to prevent monotony in training. Jumps were performed off the ground, off 12-in wooden boxes, in the forward and side directions, and in single- or double-leg stances. In general, each session consisted of equal repetitions of the following types of jumps: two-footed jumps off the ground, two-footed jumps onto and off a 12-in wooden box, two-footed side-to-side hops, and one-footed hops. Subjects performed the jumps on 2-in gymnastics mats and were instructed to jump with shoes off and to land flat-footed with approximately 30° of knee flexion. Pilot data, collected in the laboratory, generated ground reaction forces of four to five times BW for jumps and are categorized as high-intensity exercise. We define high-intensity activities as those that produce forces greater than four times BW.(17) Lower body resistance exercises (squats, lunges, and calf raises) were performed immediately after the jump exercise. Squats were performed in a wide stance to 90° of knee flexion; lunges were performed in the forward, side, and backward directions to 90° of knee flexion in the lead leg; while calf raises were performed off the toes to slightly less than 90° of plantar flexion. In general, each session consisted of three sets of 10-12 squats, six sets of 10-12 lunges (two sets in each direction), and two sets of calf raises.

Intensity for both jump and resistance exercises was increased using weighted vests and was calculated as a percentage of BW (%BW), such that each woman had the same relative intensity. Jump and resistance intensity were increased over the first 10 months at rates of 1% BW/month and 1.25% BW/month, respectively, and remained at 10% BW for jumps and 13% BW for resistance exercises during the final 2 months. Women recorded their training on individual logs kept in the exercise room and maintained records of physical activities performed outside of class.

After the conclusion of the training program, subjects in the exercise group were asked to discontinue participation in exercise training for 6 months. Women were encouraged to maintain other normal activities and dietary regimens that they had followed during the training program. Control subjects also were followed during the detraining program and were asked to maintain their usual activity and dietary habits.


Tests were performed at baseline (month 0), posttraining (month 12), and postdetraining (month 18) on all dependent measures. An additional measurement of BMD was made at the midpoint of training (month 6) to assess the time course of the bone response to training.

Anthropometric indices

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

BMD and body composition

BMD (g/cm2) of the greater trochanter, femoral neck, lumbar spine (L2-L4), and whole body was measured via dual-energy X-ray absorptiometry (DXA; Hologic QDR 1000-W, software version 4.74; Hologic, Inc., Waltham, MA, U.S.A.). Lean and fat masses were determined from whole body scans. The same individual conducted all scans and analyses. In-house CVs on a subsample of women similar to our study population are <1.0% for hip and spine measures and <1.5% for whole body and body composition measures.

Lower extremity strength

Knee extensor and hip abductor strength were assessed via isokinetic dynamometry (Kin-Com 500-H, Kin-Com, Chattex, TN, U.S.A.). All strength values were corrected for the effect of gravity on the limb in the horizontal position. This instrument has been shown to provide valid and reliable estimates of muscle strength,(18) and our in-house CV is <4%.

Subjects performed three to five trials before each maximal effort for warm-up and to ensure proper positioning. Each subject was then instructed to perform at least three maximal efforts until no further increases in strength were observed. A 1-minute rest period was allowed between successive trials and most subjects reached peak effort on the third trial. Both strength tests were performed at a rate of 30°/s to provide the optimal velocity for peak force development. For assessment of knee extensor strength, subjects were in a seated position in which full leg extension was set at 180° and the range of movement was from 85° to 150°. To measure hip abductor strength, subjects lay on their left side where horizontal placement of the tested limb parallel to the table was set at 0° and the range of movement was from −5° to 30°.

Leg power

Muscular power (W) of the legs was assessed using a modified version of the Wingate Anaerobic Power Test on a Monark bicycle ergometer (Ergometer 818E, Monark-Crescent AB, Varberg, Sweden). The test consisted of a 3- to 5-minute warm-up period of low-intensity cycling at 60-70 rpm, followed by 15 s of maximal pedaling against a resistance set at 7.5% of BW.(19) The highest value obtained during the 15-s trial was taken to reflect peak leg muscle power. The Wingate test has been shown to be a valid measure of muscular power in younger women,(20) and our in-house CV for maximal power is <4%.

Dynamic postural stability

The ability of subjects to balance themselves for 30 s on an unstable surface was tested using a stability platform (Biodex Medical Systems, Shirley, NY, U.S.A.). Values are expressed in terms of the Stability Index, a unitless measure that represents the variance of platform displacement in degrees from level. A higher number is indicative of less motion control and thus poorer postural stability. The test protocol consisted of the following three sets of trials: (1) a single positioning trial in which subjects self-selected a foot position that felt most stable on the unsteady platform (subjects kept this foot positioning for the practice and test trials); (2) two practice trials in order to negate learning effects; and (3) two test trials, which were averaged to yield the stability measure used in subsequent analyses. During the practice and test trials, subjects were instructed to keep the platform as steady as they could without using the handrails for assistance. Subjects had visual feedback in which they attempted to keep a cursor that tracked their movements in the center of a target displayed on a screen in front of them. The in-house CV for this measure is <10%.

Nutritional analyses

Macro- and micronutrient intake was assessed from 3-day self-reported food intake records. Subjects 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. 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 (ESHA Research, Salem, OR, U.S.A.).

Statistical analysis

All values are expressed as mean ± SD, except graphs 1-3 in which data are expressed as mean ± SEM. Dependent measures were examined to determine whether they met the assumptions of normality, linearity, and homogeneity of variance. Statistical analyses of the training and detraining data were conducted and interpreted separately. However, graphical representation of bone data includes all four measurement points ( Figs. 1 and 2). Training effects were assessed using a between-within subjects design, whereas detraining effects were evaluated using a within subjects design for each group separately. Initial differences between groups were determined using unpaired t-tests. For the training period, between and within group differences were evaluated by separate repeated measures analysis of covariance (ANCOVA) on the pre- and posttraining values for each dependent measure, using the baseline value as the covariate. Linear trend analyses also were conducted on 0-, 6-, and 12-month data for BMD measures to examine the time course of the bone response to exercise training. For the detraining analyses, difference scores on measured variables in the exercise group were compared with the change over detraining period using repeated measures analysis of variance (ANOVA). Control group data were analyzed in a similar manner to compare changes during the same periods. This approach best isolated the effect of detraining. All statistical analyses were performed with the SPSS statistical software program, version 9.0 (SPSS, Inc., Chicago, IL, U.S.A.) with a two-tailed significance criterion set at p = 0.05.

Figure FIG. 1.

Percent changes in BMD across training and detraining periods (mean ± SE) at the (A) greater trochanter, (B) femoral neck, (C) lumbar spine, and (D) whole body. *Exercise group significantly different from controls (p < 0.05); †change over detraining period significantly different from change over training period, within groups (p < 0.05).

Figure FIG. 2.

Percent changes in performance measures across training and detraining periods (mean ± SE) for (A) knee extensor strength, (B) hip abductor strength, (C) leg power, and (D) dynamic postural stability. *Exercise group significantly different from controls (p < 0.05); †change over detraining period significantly different from change over training period, within groups (p < 0.05).



Forty-one women participated in the training program and 33 women completed 12 months of exercise (retention, 82%) and 29 were available for follow-up after 6 months of detraining. Twenty-four women were recruited as matched controls and 20 completed the entire 18-month control period. Reasons for dropping out of the exercise program were as follows: aggravation of preexisting orthopedic problem (n = 3), relocation outside of study area (n = 1), pregnancy (n = 1), disinterest (n = 2), and time constraints (n = 1). Four of the women who completed the training program relocated outside of the study area during the detraining period and 4 women in the control group relocated during the 18-month control period. All statistical analyses were conducted on women with complete data sets for the entire 18-month period (exercisers, n = 29; controls, n = 20). Baseline data on dependent measures between those who completed the study and those who withdrew were not significantly different.

Subjects were mature premenopausal women of average height, weight, and body composition. Mean BMD values were 1-6% higher than age-matched referents supplied by the DXA manufacturer. Reference data were not available on performance measures and thus comparisons of mean values from our sample against a larger population were not possible. Daily intakes of total energy and calcium were adequate in both groups and percentages of calories consumed from carbohydrate, protein, and fat were typical of a westernized diet. There were no significant differences between exercise and control groups at baseline for anthropometric, dietary, or dependent measures (Table 1). Nutritional intake remained similar over time across groups. At the beginning of the study, no subject in either the control or the exercise group was currently participating in a structured exercise program. Except for the study exercise program, no subject in the exercise group began a structured program outside of class during training; however, 11 of the 29 exercisers participated in a community-based fitness program during the detraining period. The community exercise program was designed to promote total body muscular endurance and thus employed low-resistance, high-repetition exercises for the whole body, with no jumping. In contrast, the study exercise program emphasized high-resistance, low-repetition exercises with added jumps. No control subject participated in a structured exercise program during the 18-month study period.

Table Table 1.. Baseline Characteristics of Subjects (Means ± SD)
original image


Within groups, exercisers significantly increased BMD at the greater trochanter, femoral neck, and whole body, whereas controls significantly increased BMD only at the greater trochanter (Fig. 1; Table 2). Between groups, the increase in trochanteric BMD in the exercise group was significantly greater than the increase in controls (p = 0.005); however, a tendency for higher BMD was observed at the femoral neck (p = 0.056) and whole body (p = 0.089). Linear trend analyses of 0-, 6-, and 12-month data from the exercise group detected continuous increases over time for trochanter, femoral neck, and whole body BMD.

Table Table 2.. Changes in Anthropometric and Performance Measures After Exercise Training
original image

After 6 months of detraining, femoral neck BMD in the exercise group was significantly lower compared with posttraining values (Fig. 1; Table 3). At all bone sites, the decrease in BMD over detraining was significantly different from the gain in BMD over training. There was no significant difference between baseline and postdetraining (18 months) BMD values, suggesting that no benefit achieved from the exercise program was maintained. Over the same detraining period, controls showed a significant increase in lumbar spine BMD and no change in BMD at other sites.

Table Table 3.. Changes in Anthropometric and Performance Measures After Detraining
original image

Muscle strength, leg power, and dynamic stability

Significant increases in lower extremity strength and power over the training period were observed in exercise versus control groups ( Table 2; Fig. 2). Exercisers increased knee extensor and hip abductor strength by 17% and 27%, respectively, and leg power by 28% (Table 2). Controls increased hip abductor strength and leg power by ∼18%, roughly half the increase of the exercise group. Improvements in dynamic postural stability in exercisers were twice that of controls (24% vs. 12%), but the difference between groups was not statistically significant (p = 0.098).

Over the 6-month detraining period, the exercise group exhibited significant decreases of 8% in knee extensor and hip abductor strength and decreases of 18% in leg power ( Table 3; Fig. 2). For controls, changes in strength during the 6-month detraining period were not significantly different from changes over the first 12 months, but changes in power were significantly different between the two periods. For both exercisers and controls, dynamic stability did not change during detraining.

Body composition

Body composition improved significantly over both the training and the detraining periods in exercisers, but did not change in controls over the same time frame (Tables 2 and 3). In exercisers, lean mass increased (+1.0 kg) and fat mass decreased (−0.9) resulting in an overall decrease in percent body fat (−1.1%), whereas controls maintained body composition. During the detraining period, body fat significantly increased in the exercise group (+1.1%) but did not change in controls.


Twelve months of impact plus lower body resistance exercise significantly increased trochanteric and femoral neck BMD (2.5% and 1.5%, respectively), although only changes at the trochanter were significantly different from controls. Additionally, the training program improved lower extremity strength and power and body composition when compared with controls. Although an effect was observed on balance in the exercise group, the increase was not significantly different from controls. After a 6-month withdrawal of the exercise stimulus (detraining), BMD returned toward baseline values, as did muscle strength and power, indicating the reversibility of exercise training on the musculoskeletal system.

This study has several strengths. The exercise training program was designed specifically to reduce multiple risk factors for hip fractures by combining impact and resistance exercise into a single training protocol. Kohrt et al.(8) report that exercises that produce ground reaction forces, such as running or walking, best increase BMD whereas exercises that produce joint reaction forces, such as resistance training, best improve lean body mass and strength in postmenopausal women. They further suggest that a comprehensive program of combined exercise may yield optimal reductions in fracture and fall risk and ours is the first to examine this type of intervention in premenopausal women. Furthermore, the intervention adhered to the training principles of specificity, overload, and reversibility. To our knowledge, we are the first to include and evaluate these components of training in a study of risk factors for osteoporotic fractures. The program was specific to the lower extremities; overload via the use of weighted vests and increased ground reaction force (jumps) was applied to both bone and muscle, and the stimulus was withdrawn to study reversibility. The detraining component in our study also yields an estimate of rates of decline when exercise discontinues. Also important are the good compliance (75%), low attrition (20%), and small occurrence of minor injuries with training. The training program was enjoyed by participants, took less than 30 minutes to complete, and included simple movements that could be performed easily at home.

The limitations of this study also must be mentioned. Ours was not a randomized controlled trial, because participants initially were recruited to participate in the training program and then a matched control group was selected. Although baseline data did not differ between groups, we cannot ensure that selection bias did not occur. Additionally, the length of our detraining period was half that of our training period; thus the full extent of detraining may be underestimated; however, we are able to examine short-term effects. Furthermore, approximately one-third (11 out of 29 women) of the exercise group included in detraining analyses refused to stop exercising and continued to participate in community or home-based exercise. No study subject continued jumping exercise, but the community fitness classes consisted of resistance exercise of less intensity than the research program and put emphasis on the whole body. Exercise participants who did not fully comply with the detraining guidelines may attenuate the observed changes in functional parameters from detraining. Nevertheless, a detraining effect was clear, indicating the effectiveness of our program. When the subjects who continued to train were removed from detraining analyses, all significant tests remained. In addition, our ability to determine the relative contributions of jump or lower extremity resistance exercise to the gains in BMD is precluded by our study design. Previous studies have debated whether ground reaction or muscle forces during jump exercise contribute most to gains in BMD.(8,10) Because our study combined exercise with both impact and muscle forces, we cannot discern between the two; however, our intent was to develop a practical program to reduce fracture risk and not to determine the mechanism of the observed changes. Finally, because our program examined multiple risk factors for fracture and the effects of exercise training and detraining on them, several statistical tests were performed. This situation has the potential to inflate type I error, although our significant tests had a corresponding power of 75-90% and thus we are confident that these reflect true differences.

As previously reported, impact exercise via jumping has been shown to be osteogenic at the trochanter and the femoral neck. (9–11) The trochanter appears to respond more quickly to loading exercise and significant increases are observed within 5-12 months,(9, 10, 12) whereas the femoral neck appears to require a longer stimulus period (18 months) to note significance.(10,11) Our data corroborate those of Bassey et al.,(10) in which both trochanteric and femoral neck BMD increased in response to jump exercise, but the response at the femoral neck was slower compared with the trochanter. Additionally, we observed significant linear increases at both the femoral neck and the trochanteric regions, suggesting that lengthening the training program might further increase BMD at the femoral neck above controls while trochanteric BMD might also continue to improve. The reversal of gains in hip BMD after 6 months of detraining confirms the effectiveness of the training program. Both femoral neck BMD and trochanteric BMD approached baseline levels after half the time spent training. Thus, training withdrawal appears to have a potent effect on bone loss.

As secondary outcomes, we evaluated training and detraining effects on lumbar spine and whole body BMD. As expected, BMD remained near baseline levels at these sites in both exercisers and controls over the training and detraining periods. Attenuation of reaction forces from jump exercise by the landing surface of the foam gymnastics mat and surrounding soft tissues of the lower extremity likely prevented delivery of forces with sufficient magnitude to affect the lumbar spine. Although Heinonen et al.(11) reported significant changes at the lumbar spine from high-impact stepping routines, this type of dance exercise incorporates the trunk and upper body musculature. Resistance exercise also has been found to increase spine BMD in premenopausal women,(12,21) but successful interventions included upper body exercises, whereas our program was specific to the lower body. Furthermore, these resistance training studies supplemented subjects with 500 mg/day of elemental calcium. An interactive effect of calcium and physical activity in premenopausal women has been suggested(23,24); thus the effects of calcium on observed changes in spine or hip BMD may confound attributions of bone changes to resistance exercise. Although we observed significant differences at the lumbar spine in controls between periods, the percent changes were within our reported machine error and thus are not meaningful.

We report significant improvements in risk factors associated with falling. Specifically, lower extremity strength and power increased in the exercise group compared with the control group, and both measures are independently associated with falls.(5,6) Neither jump training study of Bassey et al.(9,10) nor Heinonen et al.(11) produced improvements in strength; however Heinonen et al.(11) did report a 20% improvement in leg power as assessed by the vertical jump test. On the other hand, resistance training studies have shown improvements in strength ranging from 14% to 34% when measured by similar methods as the present study,(12, 14, 25) but these studies did not measure leg power. The combined effects of jump plus resistance training in the present study produced increases in both strength and power of the lower extremities, similar in magnitude as reported by others. Decreases in both lower extremity strength and power among the exercise group after detraining confirm the positive training effect on these risk factors for falls. Strength and power declined nearly 50% after 6 months of detraining. If these decreases are extrapolated to the equivalent time spent training, values would decrease near or below pretraining levels. However, the effect of stimulus withdrawal on the muscular system was not as potent as it was on the skeletal system.

Dynamic stability within the exercise group increased significantly from training, and increases were nearly twice that of controls; however, changes between groups were not significantly different as were changes observed with strength and power. Because poor balance is associated with an increased risk of falls, and strength and power are related to stability, we hypothesized that the resistance component of the program would result in significant increases in dynamic stability and declines with detraining. However, stability did not decrease toward baseline in the exercise group after detraining, but rather remained near posttraining levels. Because our stability measure had a large CV and both groups showed wide variability in baseline and posttraining scores, particularly controls, the power to detect differences between groups was low, even though the improvements in the exercise group were nearly twice that of controls (24% vs. 12%). Furthermore, the plateau in balance scores after detraining suggests that a learning effect may have occurred over time. Despite efforts to account for such a phenomenon by conducting multiple trials, the plateau in scores in both groups and subjective impressions of the research staff indicate that subjects remembered adjustments to improve scores at successive testing sessions. To our knowledge, only one other intervention to evaluate exercise effects on fracture risk in premenopausal women has measured dynamic balance. Although improvements were small (0.3 s improvement [5%]), Heinonen et al.(11) reported that step and jump training significantly improved dynamic balance, as measured from a figure-of-eight run tests. The CV for this measure was not reported, and the significance value (p = 0.08) was greater than the a priori level set by the researcher (p < 0.05). The appropriateness of the figure-of-eight run as a valid measure of dynamic stability is questionable and may rather be a better measure of speed and agility, factors unrelated to falls. Our study used a valid measure of stability,(26) showing significant improvements outside of our reported CV, but had large variability and low power, thus, preventing conclusive statements regarding the effectiveness of our program to improve stability.

Body composition improved significantly in the exercise group because of both an increase in lean body mass and a decrease in fat mass. Numerous studies have shown that resistance training improves body composition by increasing lean body mass, while the increased caloric expenditure from physical activity also may reduce body fat. (27–29) When the exercise group stopped training, percent body fat returned to baseline levels, strongly confirming the ability of our training program to improve body composition.

Because advancing age and associated reductions in physical activity result in a myriad of physiological and functional declines, our training program could provide an effective and well-tolerated means for offsetting these changes. The training program was enjoyable, safe, time-efficient, and adaptable for at-home exercise. Increases in lean mass and reductions in body fat from resistance training reduce the risk of metabolic and cardiovascular disease and promote self-confidence and positive self-image.(30) Improved strength and power reduce the risk of falling and improve physical function. Equally important, increases in trochanteric bone mass from jump exercise may reduce the risk of osteoporosis in later life and the occurrence of related fractures. Trochanteric fractures carry considerably higher morbidity and mortality than fractures of the femoral neck or vertebrae(31); thus improvements at the trochanter may have a stronger influence on extending the quality and quantity of life. Particularly striking was the potent effect of exercise withdrawal on the skeleton, such that gains in BMD can be maintained only with continued training. Unless women remain committed to exercise, other preventive measures should be considered. Thus, 1 year of jump plus resistance training during premenopausal adult years may prevent or delay the development and progression of osteoporosis and other chronic diseases; however, the protective effect of exercise is dependent on the establishment of lifetime activity habits.


We gratefully acknowledge the National Aeronautics and Space Administration Graduate Student Research Program and LifeFitness, Inc., for their financial support of this project.