Prospective Ten-Month Exercise Intervention in Premenarcheal Girls: Positive Effects on Bone and Lean Mass



Enhancement of bone mineral acquisition during growth may be a useful preventive strategy against osteoporosis. The aim of this study was to explore the lean mass, strength, and bone mineral response to a 10-month, high-impact, strength-building exercise program in 71 premenarcheal girls, aged 9–10 years. Lean body mass, total body (TB), lumbar spine (LS), proximal femur (PF), and femoral neck (FN) bone mineral were measured using the Hologic QDR 2000+ bone densitometer. Strength was assessed using a grip dynamometer and the Cybex isokinetic dynamometer (Cybex II). At baseline, no significant difference in body composition, pubertal development, calcium intake, physical activity, strength, or bone mineral existed between groups. At completion, there were again no differences in height, total body mass, pubertal development, calcium intake, or external physical activity. In contrast, the exercise group gained significantly more lean mass, less body fat content, greater shoulder, knee and grip strength, and greater TB, LS, PF, and FN BMD (exercise: TB 3.5%, LS 4.8%, PF 4.5%, and FN 12.0%) compared with the controls (controls: TB 1.2%, LS 1.2%, PF 1.3%, and FN 1.7%). TB bone mineral content (BMC), LS BMC, PF BMC, FN BMC, LS bone mineral apparent density (BMAD), and FN bone area also increased at a significantly greater rate in the exercise group compared with the controls. In multiple regression analysis, change in lean mass was the primary determinant of TB, FN, PF, and LS BMD accrual. Although a large proportion of bone mineral accrual in the premenarcheal skeleton was related to growth, an osteogenic effect was associated with exercise. These results suggest that high-impact, strength building exercise is beneficial for premenarcheal strength, lean mass gains, and bone mineral acquisition.


PHYSICAL ACTIVITY HAS BEEN advocated as offering a potential means to increase and maintain bone mineral density (BMD). For hypertrophy to occur in a region, the stress to the area must be greater than the accustomed load threshold.1 Unusual strain distribution, high strains, and high strain rates are particularly osteogenic, principles which should be incorporated into exercise regimes designed to promote or maintain bone mass.1 Cross-sectional studies by Slemenda et al.2 and Teegarten et al.3 both demonstrated a significant positive association between the level of childhood physical activity and BMD. Children participating in sports generating high impact loading (greater than three times body weight) on the skeleton had greater femoral neck bone density and a trend for greater spinal bone density than children in sports producing low loading.4 Gymnasts as young as 7 years of age have also been found to have higher total body bone density compared with control subjects.5 In addition, childhood activities that preferentially stress one side of the body provide the strongest evidence that physical activity can modulate BMD during growth.6,7 However, it was not possible from these retrospective cross-sectional questionnaire-based studies to establish a direct cause-and-effect relationship between exercise and bone accrual. It has been suggested that a self-selection bias may exist for subjects with higher initial BMD or greater genetic potential for bone acquisition choosing to participate or being more suitable for participation in physical activity. Some, but not all prospective studies have demonstrated positive relationships between bone mineral measures and physical activity.8,9 However, few prospective studies have been completed in the premenarcheal population. To enhance further our knowledge of the benefits of exercise for bone mass, we need to answer a fundamental question: Are the premenarcheal years an opportune time for bone to respond to exercise?

Weight-bearing, high-impact–based activities have been shown to have the greatest influence on bone mass,10 through muscular contractions actively loading the skeleton.11 Research into the association between bone and weight-training have demonstrated positive associations between strength and BMD, and lean mass and BMD in the adult skeleton.12,13 However, few of these studies have examined the influence of weight training on lean mass changes, strength gains, and bone mineral accrual in premenarcheal children. None have studied active weight training in premenarcheal girls. Based on the importance of strength and lean mass for bone, a second question arose: Does an exercise program that ensures the greatest lean mass increase and muscular strength gains results in the greatest skeletal loading and the greatest bone acquisition?

To date, research into the benefits of exercise for lean mass and skeletal mineralization have focused on the mature skeleton. However, it is potentially the premenarcheal population that holds the most promise for promoting bone mineral accrual in response to high-impact, strength-building exercise due to its correspondence with peak growth velocity,14 peak strength gains,15 and peak bone formation and consolidation as indicated by peak levels of bone turnover markers and rapid rates of change in measured BMD.16–18 Assuming that physical activity, lean mass, and strength gains contribute to the bone mass response, greater understanding of the contribution of each of these factors to bone mineral accrual may lead to strategies to optimize peak bone mass and ultimately to reduce the risk of osteoporosis. It is our hypothesis that the premenarcheal years provide a window of opportunity for lean mass and bone to respond to exercise. We propose that the exercise-induced benefits in the premenarcheal skeleton will occur in response to high-impact, strength building exercise. The purpose of this prospective study was, therefore, to examine the lean mass, strength, and bone mineral response to a weight-bearing, strength-building exercise program in premenarcheal girls.


Research design

The research design was a longitudinal, nonrandomized, parallel group study investigating the effects of exercise on BMD over a 10-month duration, comprising three 10-week school terms. Four schools from the same ethnic, socioeconomic, and geographic area were assigned as exercise or control schools, with two schools comprising each group. The schools were matched by ethnic, socioeconomic background from statistics provided by the Ministry of Education and the Social Atlas of Melbourne.19 The exercise and control schools were allocated based on school self-selection for practical and compliance purposes. We acknowledge this as a deficiency in the method, but school consent to run the exercise program within the school curriculum was necessary prior to commencement of the study. There was no apparent bias in subject recruitment within each school, with all subjects within the one school participating in the same intervention. All subjects completed two sets of assessments at baseline and after 10 months. Prior to initiating the study, the project purposes and procedures were explained verbally and in writing to the school community, parents, and schoolgirls. Informed, written consent was obtained from the parents and schoolgirls prior to any testing, in accordance with protocols defined by the Royal Melbourne Hospital Medical Research and Ethics Committee.


At initiation of the study, all subjects were required to be between the age of 9 and 10 years, enrolled in grade 4 or 5 at school, and be at the premenarcheal stage of development. Participants were also asked to complete a medical questionnaire that evaluated their state of health and fitness. Subjects were excluded from the study if they reported any condition that might interfere with bone metabolism or strength training exercise such as diabetes, renal disease, heart disease, anemia, or chronic musculoskeletal disease. Based on these criteria, no subjects were excluded. One hundred and twenty primary schoolgirls attending the exercise and control schools were initially approached, and of the students invited to join the study, a cohort of 73 subjects agreed to participate. The primary reasons given by the girls who declined to participate in the study were that they were unable to arrive at school early to participate in the exercise program or were concerned at the radiation exposure. Seventy-one subjects completed the entire study, with two subjects departing the study due to geographical relocation away from the school. Of the 71 participants, 38 participated in the exercise program and 33 acted as controls.

Pubertal development

Sexual maturity was rated on a scale, using the criteria described by Tanner.20 Each mother in conjunction with her daughter was presented with Tanner's photographs of the stages of development of secondary sex characteristics during puberty, consisting of five stages of breast and pubic hair development, and directed to select the appropriate level. Stage 1 indicates the prepubertal stage; stage 2 indicates initial development, stages 3 and 4 indicate continued development, and stage 5 represents the adult or mature development of the characteristic. Appropriate descriptions of the five stages of breast and pubic hair development were attached. This method of sexual maturity rating correlates significantly with physician assessment of pubertal development, reporting coefficients of variations of 0.81 and 0.91, respectively, for the self/parental assessment of breast and pubic hair development in girls.21

Calcium intake

Calcium intake was assessed twice during the 10-month study. A food frequency questionnaire that was developed and validated against assessment of calcium intake by the 4-day weighed food record (r = 0.79, p < 0.001).22 The food frequency questionnaire consisted of a list of major sources of calcium, including milk, dairy products, and nondairy products. The calcium content of each food item was derived from an Australian food composition table.23 Each subject completed the food questionnaire under the guidance of a nutritionist, estimating the daily or weekly consumption of the various food items. A question relating to calcium supplementation was also included in the questionnaire. Each questionnaire was then analyzed for calcium content.

Physical activity

The physical activity level was estimated twice during the 10-month study, using a simple quantitative physical activity self-report questionnaire developed by Godin and Shephard.24 Subjects reported the number of times in an average week that they spent more than 15 minutes in activities that were classified as mild (3 METs), moderate (5 METs), or strenuous (9 METs). METs, or metabolic equivalents, represent multiples above the resting metabolic rate. A total score was derived by multiplying the frequency of each category by the MET value, and those products were summed. To simplify the questionnaire, activities were scored only if they totaled 15 minutes or more in an intensity category. Each questionnaire was administered with the assistance of a research assistant. A test-retest reliability coefficient of 0.81 has been reported.24 The questionnaire also included the number of minutes per week each child participated in school physical education classes and the total minutes per week engaged in school and club organized sports.

Physical activity training program

The exercise intervention program incorporated a 30-minute physical activity session, three times per week, under the supervision of a single qualified physical education teacher (Joanne Gibbs). This physical education teacher was responsible for coordinating the running of all activities within the program and taking attendance rolls at each session as a measure of compliance. The exercise program was designed to correspond to the three school terms. Exercise training commenced in February and finished in September. Training was interrupted at the end of each term by a 2-week vacation, where continued exercise was encouraged but not monitored. The exercise program included a variety of vigorous, high-impact aerobic workouts, such as aerobics, modified soccer and Australian Rules football, step aerobics, bush dance, skipping, ball games, modern dance, and weight training.

As part of the program, a 20-station weight-bearing, strength-building circuit was developed by a strength and conditioning trainer. Each circuit session incorporated 10 exercises designed to stress all body parts, including arms (biceps, triceps), chest (pectoralis major), back (latissimus dorsi, trapezius), trunk, shoulder (deltoids), stomach (rectus abdominus), and legs (quadriceps, hamstrings, gastrocnemius, soleous). Each weight-lifting exercise was carefully taught to each girl, with correct technique carefully monitored throughout each weight circuit. The circuit required the participants to alternate between 10 exercise stations, working for approximately 1 minute at each station. The initial circuit was performed for one set of 10 repetitions, with a progressive increase to three sets, in order to achieve a training overload and physiological improvement. The weight training component operated for 10 weeks duration. The exercise stations incorporated the use of 2-kg dumbbells, elastic band resistance, and movement employing the subject's own body weight.

The control subjects were encouraged not to change their physical activity patterns. To monitor physical activity patterns, a physical activity diary was completed by the control subjects at two occasions, once at baseline and once at study completion. Cross-contamination of the intervention between the exercise and control schools was thought unlikely due to the geographic separation of the exercise and control schools.

Strength assessment

Strength testing occurred within 1 week prior to and following the completion of the exercise training program. Pre- and postpeak torque measurements were made on the flexor and extensor muscles of the shoulder and knee using the Cybex Isokinetic dynamometer (Cybex II). Each subject was familiarized with the testing apparatus and protocol prior to test administration and performed one practice trial. To ensure consistency, the testing was monitored by the same technician for all subjects and each test. Both flexion and extension was recorded during concentric and eccentric activity at an angular velocity of 60°/s. During the knee strength test, the subject was seated in an armless chair and restrained via belts to prevent extraneous movement of the body. The axis of the dynamometer resistance arm was aligned with the lateral epicondyle of the femur. Concentric leg flexion strength was measured through an 80° arc, ranging from the start position of 90° of knee flexion to the end position at 180°. Subjects performed one set of three repetitions, from which peak torque was noted as the parameter for maximal strength. The strength values were recorded in foot-pounds (as given by the Cybex testing device), with a conversion made to Nm (1 ft-lb = 1.36 Nm). Pre- and postgrip strength of the dominant and nondominant hand was measured using a Smedley grip strength dynamometer. Each subject performed three trials, with the best score entered for data analysis. Results were reported in kilograms. The precision of the Cybex machine for knee flexion and extension using the above protocol has been estimated to be 7.5% and 5.5%, respectively.25 Precision of the Cybex for pediatric shoulder flexion and extension has been estimated to be 5.3–5.8%.26

Bone mineral density and body composition measurement

The Hologic QDR-2000+ dual-energy X-ray absorptiometer was used to assess soft tissue composition and bone density. Each subject underwent a whole body scan, from which soft tissue composition was determined. The analysis provided mass in grams of body fat, lean tissue, total body tissue, and percentage body fat. The reported coefficient of variation in our laboratory was 0.8% for lean mass and 1.3% for fat mass in adult subjects. Bone mass was measured at the total proximal femur (PF), neck of femur (FN), lumbar spine (LS), and total body (TB) at baseline and 10 months using dual energy X-ray absorptiometry (DEXA, QDR 2000+, Hologic Inc., Waltham, MA, U.S.A., software version 6.3). The standard Hologic protocol for positioning the TB, LS, PF, and FN sites was utilized, including a constant FN width when assessing bone area. Bone mineral results were expressed as BMD (g/cm2), bone mineral content (BMC, g), and bone area (area, cm2). The BMD data also were adjusted for the size of the bone, expressing the results as bone mineral apparent density (BMAD). BMAD is equal to the BMC divided by a derived bone reference volume. For the lumbar spine, Carter et al.27 used the square root of the projected area Ap of (L2–L4), calculated as BMAD = BMC/(Ap)1.5. For the femoral neck, Katzman et al.28 used the area squared, calculated as BMAD = BMC/A2. BMD assessments were made 1 week prior to commencement and at the completion of the exercise training program. All scans were conducted and analyzed at the Essendon Osteoporosis Centre (a campus of the Royal Melbourne Hospital). All scans were conducted by the same radiographer, who was blinded as to the subject's group. The in vivo precision of DEXA in our laboratory is approximately 1.0% for total body BMD, 0.7% for leg BMD, 1.3% for lumbar spine BMD, 1.0% for the femoral neck, 0.7% for the proximal femur BMD, and 0.6% for total body BMC. The coefficient of variation provided by Carter et al.27 for LS BMAD was 11.4% for the lumbar spine BMAD.

Statistical analysis

Statistical analysis was carried out using SPSS for windows (version 6.0, SPSS Inc., Chicago, IL, U.S.A.). Student t-tests were used at baseline to test for significant differences in body composition, strength, and bone mass between the exercise and control groups. Absolute and percent changes from baseline were calculated for height, total body mass, pubertal development, muscular strength, bone mineral accrual, and independent t-tests used to compare the significance of these changes between groups. Partial correlation, adjusting for height, and multiple regressions were performed with changes in bone density as the dependent variables and changes in strength and anthropometric variables entered individually as determinants. An α level of 0.05 was accepted for significance of all statistical procedures.


Of the initial cohort of subjects, all controls completed the study, with only two exercise subjects from the initial group of 40 dropping out. This represented an attrition rate of 2.7%. Reasons for not completing the study were, in both instances, departure from the school. Physical activity participation compliance for the exercise group, defined as the number of exercise sessions attended, was 96% midway through the study and 92% at the completion, indicating that the exercise program was well accepted by all participants. Baseline subject characteristics of the 38 exercise and 33 control were similar for age, height, total body mass, body fat content, total body lean mass, leg lean mass, arm lean mass, pubertal development stage, calcium intake, school sport, school physical education class time, and physical activity external to the exercise intervention program as shown in Table 1.

original image

At the completion of the study, there were no significant differences in the rate of change in height, total mass, pubertal stage, calcium intake, and external physical activity levels, with both groups following a similar growth and maturation process. No schoolgirl commenced menstruation during the study. However, significant differences existed for body composition changes, strength changes, and bone mineral accrual, which will be discussed in turn. The exercise group gained greater total lean mass, trunk lean mass, and leg lean mass compared with the control subjects (p < 0.01). In contrast, the control group gained significantly more total body fat mass compared with the exercise group (p < 0.04) (Table 1).

The rates of change in shoulder extension, shoulder flexion, knee extension, and nondominant grip strength increase exceeded the change for the controls by 26, 33, 7, and 20%, respectively.

Baseline BMD measurements were similar for the exercise and control groups at the commencement of the study. At completion, the exercise group accrued significantly greater TB, LS, leg, arm, pelvic, FN, and PF BMD compared with the control subjects as illustrated in Table 2. The absolute change in TB, LS, and FN BMD are illustrated in Fig. 1.

Figure FIG. 1.

Absolute change in BMD across 10 months (mean ± SD).

original image

TB BMC, LS BMC, PF BMC, FN BMC, and LS BMAD change over the 10 months were also significantly greater in the exercise subjects, following a similar trend to the BMD accrual (Table 2). No difference was observed in TB, PF, or LS bone area; however, FN bone area increased significantly more in the exercise group (Table 2).

Exercise was associated with greater bone mineral accrual, an effect that was independent of changing height and total body mass. TB BMD (p < 0.01), TB BMC (p < 0.05), LS BMD (p < 0.05), PF BMD (p < 0.05), and FN BMC (p < 0.05) accrual were significantly greater in the exercise group, controlling for alterations in height and total body mass. No difference was observed in LS BMC (p < 0.76) and FN BMD (p < 0.71) accrual between groups, controlling for height and total body mass change.

To determine the factors predictive of bone accrual, changes in regional body composition and strength were correlated with bone mineral accrual. In general, increased lean mass was positively associated with increased TB, LS, PF, and FN BMD and TB, LS, and PF BMC as shown in Table 3. For the exercise group, increased height, total body mass, total body lean mass, regional lean mass, and shoulder strength were positively associated with TB BMD and TB BMC accrual. At the lumbar spine, gains in total body lean mass and trunk lean mass were associated with increased LS BMD and LS BMC. At the proximal femur, increases in total lean mass were associated with increased PF BMD and PF BMC. At the femoral neck, increases in total body lean mass and leg lean mass were associated with FN BMD and FN BMC accrual. Fat mass and strength changes were not associated with bone accrual at any site. For the control group, increased height and knee extension strength were positively associated with TB BMD and TB BMC accrual. At the lumbar spine, increased total body mass and total body fat content were positively associated with LS BMC accrual. Again, strength changes were not a significant predictors of bone accrual at any site (Table 3).

original image

To determine the factors predictive of changes in bone mass during the premenarcheal years, alterations in height, weight, lean mass, fat mass, pubertal stage, calcium intake, and strength variables were regressed with bone accrual. For the exercise group, change in lean body mass was an independent predictor of change in TB, LS, PF, FN BMD, and TB BMC, accounting for 10–58% of the variance in bone accrual. Change in trunk lean mass was an independent predictor of change in LS BMC, accounting for 22% of the variance in bone accrual, respectively. For the control group, multiple regression analysis did not identify significant predictors of TB BMD, TB BMC, LS BMD, FN BMD, or FN BMC accrual. Increased fat mass was the only independent predictor of bone mass change, accounting for 13% of the change in LS BMC.


This study provided direct evidence that exercise enhances bone accrual in the premenarcheal skeleton. If exercise can promote greater bone accrual and subsequent peak bone mass, exercise may be an important contributor to the prevention of osteoporotic fracture in later years. This study also provided evidence of an association between gains in site-specific lean mass and bone accrual, suggesting that exercise resulting in the greatest increase in lean mass may best enhance peak bone mass. Although previous research has suggested that exercise and weight training programs provide beneficial effects for lean mass and BMD,29 our study was one of the few prospective studies to demonstrate benefits for the premenarcheal skeleton.

In response to high-impact, strength-building exercise, total body, lumbar spine, femoral neck, pelvic, leg, and arm BMD increased at a significantly greater rate in the exercise group compared with the controls. Because BMD is often used as an important predictor of fracture risk,30 the report of greater BMD accrual in response to exercise was encouraging.

In addition to the greater BMD accrual in the exercise group, this study also demonstrated greater total body, lumbar spine, and femoral neck BMC accrual in response to the exercise program. Changes in BMC represent the total growth effects at physes throughout the skeleton plus the effects of modeling on the periosteal surfaces and remodeling on other surfaces.31 Our findings suggest exercise may have enhanced this process.

Although large increases in BMD and BMC were reported in the premenarcheal exercise group, a proportion of the bone accrual was attributable to bone growth. To minimize the confounding effect of changing bone dimensions associated with growth, we also interpreted bone accrual in terms of BMAD, reducing the dependence of bone accrual on bone size and changing dimensions.26,27 The critical element in calculating BMAD was estimating a volumetric BMD, derived from a scaling equation, by which volume was estimated from bone area measurement and the estimated bone thickness. Importantly, BMAD remained greater in the exercise group, but the difference did not reach significance at the femoral neck. The clinical importance of BMAD measurements is yet to be determined as a diagnostic tool in osteoporotic fracture risk prediction. The lack of significant difference in femoral neck BMAD accrual was possibly explained by the fact that the femoral neck bone area also increased at a greater rate in the exercise group. This finding was suggestive that exercise stimulated the bone modeling process at the femoral neck, expanding the bone size to produce a larger, presumably stronger bone. Thus, although some of the bone accrual in the premenarcheal subjects was due to bone dimensional changes accompanying growth, exercise appeared to have an influence on both bone mass and architecture.

A unique finding to this study was the increase in femoral neck bone area. Assuming that the femoral neck width remained within the default width set at 16 pixels or 1.6 cm (Hologic protocol), the increase in FN bone area may reflect an increase in the length of the femoral neck. The finding was not apparent at the lumbar spine or total body, indicating a site-specific effect and not a global response. This possible increased bone width has major clinical significance for hip fracture reduction if these skeletal benefits are maintained throughout life.

Implications from these findings are that the premenarcheal years may provide a window of opportunity for the bone modeling and remodeling process to respond to mechanical loading. Our data, and that of others, suggest that exercise-induced bone modeling is greater in the young skeleton, corresponding to a stage of development when the bone modeling process is most active.18 Blood and urine concentrations of all biochemical markers are at their highest levels during growth, starting to decline with the onset of menarche through to the cessation of longitudinal bone growth and consolidation.18 Greater exercise-induced bone modeling in the young skeleton also corresponds to a stage of development when prostaglandins and growth hormone concentrations are at their highest levels6 and when estrogen and testosterone concentrations are rising in preparation for menarche and likely to enhance the process of bone modeling.32 The adaptation of bone to increased mechanical loading has already been demonstrated in animal models, where the effect of mechanical loading on the growing bones of rats resulted in greater bone acquisition and increased bone size compared with adult counterparts.33 In humans, the skeletal benefits from tennis and squash training were twice as great if the loading commenced prior to menarche rather than after,34 suggesting that the human response to mechanical loading is more sensitive prior to the attainment of skeletal maturity. Whether these osteogenic effects are maintained into the adult years remains a key issue and needs to be addressed through on-going prospective monitoring.

Although other prospective studies have demonstrated increased BMD in response to exercise, the magnitude of the changes in the premenarcheal population was very encouraging. For example, the exercise group accrued 5.5% greater total body BMC, 3.6% greater lumbar spine BMD, and 10.3% greater femoral neck BMD above the accrual rate of the controls. This net bone accrual rate was greater than the changes reported in adult populations. Prospective studies in young women, aged 20–25, reported a 1.3% total increase in lumbar spine BMD in response to an aerobics and weight training program,35 a 1.3% increase in lumbar spine BMD in runners and no change in femoral neck BMD,29 a 1.2% increase in lumbar spine BMD in weight lifters and no change in femoral neck BMD,29 and a 2.9% increase in lumbar spine BMD in novice college rowers exposed to high mechanical loading at the spine.36 Similarly, premenopausal women only demonstrated a 0.8% increase in lumbar spine BMD in response to a weight training intervention, despite reported strength increases.37 Even less encouraging was the study by Rockwell et al.,38 who found a 4% decrease in lumbar spine BMD in an exercise group, in response to a 9-month weight training program. Although it is difficult to compare studies directly, due to the varying nature of the exercise interventions, the general consensus suggests exercise provides an osteogenic effect on the skeleton.

Another notable finding from our research was the documentation of an association between increased lean mass and bone accrual, with the changes in total body and regional lean mass emerging as the most robust determinant of BMD and BMC acquisition in the premenarcheal skeleton. It was plausible that the increased rate of bone accrual may have been a response to the higher mechanical loading generated by the greater lean mass. Regionally, a significant positive correlation was observed between increased trunk lean mass and lumbar spine bone mineral and increased leg lean and femoral neck bone density, suggesting a site-specific relationship between increased lean mass and increases in BMD. Doyle et al.13 were some of the first to explore the relationship between bone and associated musculature in human cadavers. In comparing the weight of the psoas muscle and the third lumbar vertebra, a high correlation (r = 0.72) was observed. Doyle et al.13 hypothesized that the weight of the muscle reflects the force that it is capable of exerting on bone. The relationship between lean mass and BMD was subsequently demonstrated in young girls aged 8 to 16 years,7 young twins aged 10 to 26 years,39 young women approaching peak BMD,29,40,41 premenopausal women,40 and postmenopausal women.42 Although data suggest a possible quantitative relationship between the amount of bone and the mass of the attached musculature, lean mass accounted for an average of only 20% of the variance in bone mineral acquisition in the premenarcheal girls, leaving a substantial portion unexplained. Such findings were suggestive that other factors influence bone accrual, including hereditary factors,43 lifestyle history, hormones,44 and nutritional including calcium intake,45 all of which should be taken into account when attempting to predict an individual's bone mineral status and responsiveness to interventions.

The premenarcheal years appear to be an opportune time to gain osteogenic benefits from exercise, corresponding to a time of greatest bone modeling, increasing concentrations of estrogen, and high growth hormone concentrations, and possibly an ideal time to establish positive attitudes toward exercise participation. The establishment of an optimum level of bone during the growth years is an important consideration in terms of lifelong skeletal adequacy. Continuation of the exercise program will further benefit the postmenarcheal skeleton, maintaining the bone mass achieved during growth. The results of this study suggest that by increasing the magnitude of the mechanical loading on the bone through increased lean mass and engaging in high-impact exercise, it was possible to stimulate a greater increase in bone mineral accrual. Although a large proportion of the observed changes in bone mineral were associated with growth, the results suggest that bone accrual may be enhanced by mechanical loading. If the 10% net increase in femoral neck BMD in the exercise group is sustained into adulthood, the higher peak bone mass translates into a substantial reduction in relative hip fracture risk.45 Although our results showed a promising trend for the benefits of exercise on bone accrual, continued monitoring throughout growth and adulthood is recommended to assess the implications for the establishment and maintenance of greater peak bone mass. A longer intervention period may also yield larger beneficial effects.


This research was conducted as a joint project between The University of Melbourne and Victoria University. The research team acknowledges the work of Joanne Gibbs (implementation of the exercise program), Tony Hewitt (design of the weight training program), Nikki McGrath (Essendon Osteoporosis Centre; coordination of bone densitometry scans), and Warren Payne (project design).