Jumping Improves Hip and Lumbar Spine Bone Mass in Prepubescent Children: A Randomized Controlled Trial



Physical activity during childhood is advocated as one strategy for enhancing peak bone mass (bone mineral content [BMC]) as a means to reduce osteoporosis-related fractures. Thus, we investigated the effects of high-intensity jumping on hip and lumbar spine bone mass in children. Eighty-nine prepubescent children between the ages of 5.9 and 9.8 years were randomized into a jumping (n = 25 boys and n = 20 girls) or control group (n = 26 boys and n = 18 girls). Both groups participated in the 7-month exercise intervention during the school day three times per week. The jumping group performed 100, two-footed jumps off 61-cm boxes each session, while the control group performed nonimpact stretching exercises. BMC (g), bone area (BA; cm2), and bone mineral density (BMD; g/cm2) of the left proximal femoral neck and lumbar spine (L1-L4) were assessed by dual-energy X-ray absorptiometry (DXA; Hologic QDR/4500-A). Peak ground reaction forces were calculated across 100, two-footed jumps from a 61-cm box. In addition, anthropometric characteristics (height, weight, and body fat), physical activity, and dietary calcium intake were assessed. At baseline there were no differences between groups for anthropometric characteristics, dietary calcium intake, or bone variables. After 7 months, jumpers and controls had similar increases in height, weight, and body fat. Using repeated measures analysis of covariance (ANCOVA; covariates, initial age and bone values, and changes in height and weight) for BMC, the primary outcome variable, jumpers had significantly greater 7-month changes at the femoral neck and lumbar spine than controls (4.5% and 3.1%, respectively). In repeated measures ANCOVA of secondary outcomes (BMD and BA), BMD at the lumbar spine was significantly greater in jumpers than in controls (2.0%) and approached statistical significance at the femoral neck (1.4%; p = 0.085). For BA, jumpers had significantly greater increases at the femoral neck area than controls (2.9%) but were not different at the spine. Our data indicate that jumping at ground reaction forces of eight times body weight is a safe, effective, and simple method of improving bone mass at the hip and spine in children. This program could be easily incorporated into physical education classes.


OSTEOPOROSIS IS a disease of crisis proportions. Low bone mass is a major contributing factor associated with osteoporosis-related fractures.(1–3) The most effective way to prevent osteoporosis may be to increase bone mineral content (BMC) during childhood, thereby developing a stronger skeletal foundation to offset age-related bone loss.(4–5)

Physical activity is advocated as one strategy for enhancing peak bone mass during childhood as a means to reduce osteoporosis-related fractures.(4–5) Both cross-sectional and longitudinal investigations have documented the positive effect of physical activity on growing bones, reporting higher bone mass in active children compared with nonactive children.(6) More specifically, children engaged in high-intensity weight-bearing activities such as gymnastics and ballet(7–11) have higher bone mass when compared with children involved in low-intensity weight-bearing activities such as walking and swimming.(8,10–11) Evidence that physical activity may be an effective strategy for the prevention of osteoporosis also may be inferred from cross-sectional investigations of retired athletes, showing higher bone mass with a history of childhood weight-bearing physical activity.(9,12-15) Therefore, the development of “bone loading” exercise programs targeted at increasing bone mass during childhood has important implications as a prevention strategy for osteoporosis. The ideal program is one that could easily and safely be incorporated into a physical education curriculum.

Based on the theory that high-intensity forces elicit greater changes in bone mass than low- to moderate-intensity forces,(16–17) we used gymnastics as a model to develop a highly specific jumping program and tested its efficacy in prepubescent children. The program was designed to produce ground reaction forces of eight times body weight when jumping from a 61-cm-high box, less than gymnastics (10–15 times body weight), but higher than those reported from running (2-3 times body weight).(18) We define high intensity as forces greater than four times body weight, moderate intensity as two to four times body weight, and low intensity as less than two times body weight.(19) Thus, the aim of this investigation was to examine the effects of a high-intensity jumping program on hip and lumbar spine BMC in prepubescent children.


Design and participants

The exercise program was incorporated into the curriculum of an elementary school in Corvallis, OR. This school was selected based on the large number of children enrolled in the primary classrooms (kindergarten through third grades). Those children given parental permission to participate (testing measurements and exercise intervention) were assigned randomly by gender to either a jumping or a control group on their first visit to the laboratory.

One hundred and twenty children were enrolled in five primary classrooms (approximately 25 children per classroom; Fig. 1). One hundred of these children (51 boys and 49 girls) were given parental permission to participate. The parent of each child completed a standard health and physical activity questionnaire before participation to identify inclusion into the study. Exclusion factors included disorders or medications known to affect bone metabolism, thyroid disease, diabetes, chronic diseases; orthopedic problems that may limit training and testing; body weight that exceeds 20% of the recommended weight for height and age; and a change in Tanner stage from baseline. One boy exceeded 20% of the recommended weight for height and age and was excluded. Seven children did not return for posttesting, including three jumpers (1 boy and 2 girls) and 4 controls (2 boys and 2 girls). Of these children, 4 moved, 2 parents became concerned with X-ray exposure, and 1 parent did not have time for testing.

Figure FIG. 1.

Participant profile.

Eighty-nine children (45 jumpers and 44 controls) completed the intervention and had ethnic backgrounds as follows: 87 white, 1 Asian girl, and 1 white-Hispanic girl. The study was approved by the Oregon State University Institutional Review Board and the Oregon Board of Radiology. Parents of all children gave written informed consent before participation. All testing measurements were conducted at the Bone Research Laboratory over a 2-week period at baseline and at the completion of the 7-month exercise intervention.

Anthropometric measurements and secondary sexual characteristics

Height and weight were measured in light indoor clothing without shoes. Height was recorded to the nearest 0.1 cm using a wall-mounted stadiometer (model S-220; Seca, Hanover, MD, USA) and weight was recorded to the nearest 0.1 kg using a Seca electronic weighing scale (model #770; Seca). Body fat was estimated using gender-specific prediction equations formulated by Williams and coworkers.(20) Two anatomical sites (triceps and subscapular) were measured on the participants right side using Lange (Cambridge Scientific Industries, Inc., Cambridge, MD, USA) skinfold calipers (precision error = 2%, based on a subsample of 20 children randomly chosen from our population). The same technician performed all anthropometric measurements.

Tanner stages were used to assess sexual maturation.(21) Parents were given line drawings and written explanations of each developmental stage, using pubic hair in boys and both pubic hair and breast development in girls. A researcher knowledgeable with the Tanner stage criteria was available to answer questions.

Bone measurements

BMC (g), bone area (BA; cm2), and bone mineral density (BMD; g/cm2) of the left proximal femoral neck and lumbar spine (L1-L4) were evaluated by dual-energy X-ray absorptiometry (Hologic QDR/4500-A; Hologic, Inc., Waltham, MA, USA). Bone measurements of the hip and lumbar spine have an in-house precision error of 1–1.5% based on adult scans. It was not possible to develop precision error for children in our laboratory because of increased X-ray exposure. Hip scans were performed using a positioning apparatus that held the left leg in an internally rotated position of 30°. Because of the small femoral neck size of the children in our study, the default femoral neck box of 14 pixels was reduced to ensure that the head of the femur was not included in the analysis. The femoral neck region of interest established for each child remained constant for pre- and postanalyses.(22) Lumbar spine scans were performed with the child positioned in a supine position with the knees at 90°of flexion, elevated on a semisoft box provided by Hologic, Inc. Low-density threshold spine software was used to analyze all lumbar spine scans.(23)

Ground reaction forces

Peak ground reaction forces were calculated in a subsample of volunteers (n = 16 jumpers; n = 8 stretchers) at posttesting. Children from both groups were measured to obtain data on children with and without jumping experience. Children were asked to perform 100 two-footed jumps onto a 40 cm × 60 cm force plate (model 9281B; Kistler Instrument Corp., Amherst, NY, USA) from a height of 61 cm over a period of approximately 15 minutes. Verbal and visual instructions for how to perform the jumping exercises were given to each child. After each jump trial was completed, participants returned to the 61-cm box by first stepping onto a 20-cm box and then to the 61-cm box. Each subject was allowed to proceed through the 100 trials at his/her own pace. A trial was considered acceptable when both feet made complete contact with the force plate. An ideal assessment of the kinetics of each leg on landing would require two force plates or landing on the force plate with one leg on the force plate and one leg off. However, for the purpose of this study it was assumed that each leg was subjected to exactly half of the total measured ground reaction force. It was thought that the children would not perform the jumps naturally if they were required to target half of the force plate. An average value for the peak ground reaction force was calculated across 100 trials.

Physical activity and calcium intake

Physical activity was assessed by parent and child using a self-report physical activity questionnaire developed for older children and adolescents.(6) Baseline physical activity data pertain to activities a child engaged in during the previous school year, and posttesting physical activity data pertain to activities a child engaged in during the intervention school year. Because of difficulties in obtaining accurate information on the amount of time spent participating in various nonorganized physical activities, we report data on the mode, frequency, and duration of organized physical activities (i.e., team sports and lessons) and the number of children who reported engaging in various nonorganized activities. Information obtained from this questionnaire was verified by the same researcher at baseline and postintervention.

Dietary calcium intake was obtained using the Harvard Medical School Youth Diet Survey developed for older children and adolescents between the ages of 9 and 18 years.(24) This questionnaire was designed to be self-administered; however, because of the age of the children in our study, a parent of each child was responsible for completing this questionnaire with his/her child.(25) A researcher knowledgeable with the food survey was available to answer questions regarding the classification of foods and food serving sizes. Food models were provided to aid in estimating serving sizes. Completed food surveys were sent to Harvard Medical School for analysis.

Exercise intervention

The exercise intervention was conducted from October 1998 to May 1999, during which time 3 weeks were taken off for winter break and 1 week was taken off for spring break. All children were involved in regularly scheduled physical education classes once a week for 30 minutes, taught by a physical education teacher at the elementary school. Our exercise program was incorporated into the regular school schedule, three times per week for 20 minutes, and took place on separate days from the regularly scheduled physical education classes. Because the exercise program was included as a regular classroom activity, all 120 children enrolled in the primary classrooms were required to participate in the exercise program. A teacher from each primary classroom attended the exercise classes to monitor behavior and participation.

A total of 73 exercise sessions took place during the 7-month exercise intervention. The jumping and stretching classes were led by a researcher from our laboratory, in addition to four instructors trained in teaching elementary school-aged children. The children were exposed to the same exercise instructors for the entire duration of the exercise program. The general format for each exercise class included a 5-minute warm-up, 10 minutes of either jumping or stretching, and a 5-minute cool down. Compliance was maintained by providing the children with game days once a month, intrinsic and extrinsic motivation, and classroom awards. Children in both groups were asked not to perform the jumping exercises outside of the regularly scheduled intervention classes. A researcher from both exercise groups maintained a record of attendance, lesson plans, injuries, illnesses, and the number of jumps and/or stretches that were completed each session. Attendance was calculated based on the total number of classes completed by each participant, divided by the total number of exercise classes.

Jumping group program:

Jumps were performed in a unilateral direction off of 61-cm-high boxes (Fig. 2). A 20-cm-high box was placed in front of the 61-cm-high box as a step onto the higher box. Children were taught to jump off the box with straight posture and land flatfooted with the knees slightly bent. Jumping classes took place in the school gymnasium on a wooden floor and all children were required to wear shoes when jumping. The first week (three sessions) was spent learning correct, safe jumping techniques without using the boxes. By the second week, the children progressed to the 61-cm-high boxes, using the 20-cm-high boxes as a step. Children progressed from 50 to 80 jumps per day over the next 12 sessions, increasing 10 jumps per week. At the start of the fifth week, 100 jumps per day were performed for the remaining 58 sessions. To provide variety, boxes were arranged in rows, circles, and other patterns using between 10 and 20 boxes. For example, 20 boxes would be arranged in the shape of a triangle and the children would perform five laps around the triangle, totaling 100 jumps. After the completion of each jump, children would walk/skip/run to the next box and then step up onto the next box before jumping. Children did not jump up onto the boxes. To ensure that an accurate number of jumps were completed, the children placed straws, beanbags, and other objects into large baskets after completing each jump.

Figure FIG. 2.

Pictorial representation of jumping exercises. Children performed 100 two-footed drop landings off of a 61-cm-high box onto a wooden floor three times per week.

Control group program:

The control group had equivalent contact time with their instructors and performed nonimpact stretching exercises while their classmates were jumping. Six to eight upper and lower body exercises were completed each session. Stretches were held for 15–60 s, and children performed one to two repetitions of each exercise.

Statistical analyses

All data were analyzed using SPSS version 9.0 (SPSS, Chicago, IL, USA). Univariate analysis of variance (ANOVA) was used to examine baseline differences between the jumping and control group for all anthropometric characteristics, dietary calcium intake, and bone variables (BMC, BA, and BMD of the femoral neck and lumbar spine). Two (group) by 2 (time) repeated measures analysis of covariance (ANCOVA) analyses were performed to evaluate the effects of the intervention for each bone variable. Absolute difference scores (postintervention value − baseline value) were entered as the dependent measure, group was entered as the fixed variable and initial age and bone values, and height and weight change values were entered as covariates. Rationale for using covariates is based on literature identifying age, height, and weight as influential factors on the growing skeleton.(6,26–30) Because BMD does not accurately correct for changing bone geometry in the growing skeleton, BMC was the primary outcome variable because it reflects both the material and the geometric properties.(31) Significance level is reported as an α-level at or below 0.05 and all data and graphs are presented as means ± SEM.


Subject characteristics

Jumpers and controls were similar at baseline and at the completion of the intervention for all anthropometric characteristics (Table 1). All children were classified as Tanner stage I at baseline and posttesting. Tanner stage I is noted as prepubescent, with no signs of secondary sexual characteristics. One girl was excluded from the final analyses because of a change in Tanner stage at baseline (stage I) to postintervention (stage III). Inclusion of her values did not influence the overall effects of the intervention.

Table Table 1.. Baseline and Postintervention Anthropometric Characteristics by Group
original image

Bone measurements

At baseline, jumpers and controls had similar values for BMC, BA, and BMD at the femoral neck and lumbar spine (Table 2). After 7 months of exercise, in repeated measures ANCOVA (covariates, initial age and bone values and height and weight change values), jumpers had significantly greater changes in femoral neck BMC (0.150 ± 0.016 vs. 0.066 ± 0.016 for jumpers and controls, respectively; p < 0.001) and femoral neck BA (0.161 ± 0.014 vs. 0.083 ± 0.014 for jumpers and controls, respectively; p < 0.001). Both jumpers and controls had similar changes in femoral neck BMD (0.022 ± 0.003 vs. 0.014 ± 0.003, respectively; p > 0.05; Table 2; Fig. 3). Jumpers had significantly greater changes in lumbar spine BMC (1.956 ± 0.184 vs. 1.26 ± 0.19, respectively; p < 0.05) and lumbar spine BMD (0.021 ± 0.003 vs. 0.010 ± 0.003, respectively; p < 0.01) than controls. There were no significant changes between groups in lumbar spine BA (jumpers 2.01 ± 0.240 vs. controls 1.57 ± 0.243; p > 0.05; Table 2; Fig. 3).

Figure FIG. 3.

Seven-month changes in femoral neck and lumbar spine BMC were significantly greater in jumpers (n = 45, black bar) than controls (n = 44, white bar). Values reported as percent change (%), mean ± SEM.

Table Table 2.. Baseline and Postintervention Values by Group for Femoral Neck and Lumbar Spine BMC, BA, and BMD
original image

Exercise intervention compliance and injury

There was an overall attendance (compliance) of 96% (range of 86–100%). Class absences were caused by illness, injuries (not associated with the exercise intervention), vacation, and school-related activities. There was no correlation between the number of classes attended and the bone response to the exercise intervention.

No major injuries occurred during the exercise intervention in either the jumping or the control group; however, there were occasional minor abrasions on the hands and shins in the jumping group because of bumping into the sides of the wooden boxes. No participants discontinued the exercise program because of pain or injury from the jumping or stretching exercises. At the start of the intervention some of children told the primary instructor that the jumping exercises made their feet/knees sore; however, as the children adapted to performing the jumping exercises the number of complaints associated with foot and knee pain was reduced. There were no reports of pain or discomfort in the lower back region, hips, and shins. At the completion of the intervention, children in the jumping group indicated that their legs felt stronger and the exercises had become easier to perform.

Ground reaction forces

Average ground reaction forces for 100 trials was 8.8 ± 0.9 times body weight for jumpers and 8.6 ± 1.05 times body weight for controls.

Physical activity and calcium intake

Both groups reported participating in similar amounts and types of activities during the exercise intervention. Forty-nine children reportedly engaged in the following organized team sports: soccer (15 jumpers and 17 controls), baseball (7 jumpers and 7 controls), gymnastics (1 jumper and 4 controls), basketball (4 jumpers and 3 controls), football (1 jumper and 1 control), swimming (1 jumper and 1 control), and roller-hockey (1 control). Organized team sport seasons were 8–10 weeks in duration, with one to three practices/games per week. During the 7-month intervention, 30 children (16 jumpers and 14 controls) participated in one team sport, 16 children (7 jumpers and 9 controls) participated in two team sports, and 3 children (1 jumper and 2 controls) participated in three team sports. Six children (4 jumpers and 2 controls) started a new team sport during the study that they had not engaged in the previous school year. Of these children, 2 jumpers had not participated in any organized sports in the previous school year. In addition to organized team sports, 70 children (31 jumpers and 39 controls) reported engaging in running activities/games after school and at recess, 54 children (29 jumpers, 25 controls) reported engaging in cycling, and 26 children (15 jumpers and 11 controls) reported engaging in swimming during the 7-month intervention. Only one child (jumper) reported no participation in physical activities.

Calcium intake was not significantly different between groups based on dietary information derived from the Harvard Youth Food Frequency Questionnaire. Dietary calcium intake for jumpers was 1286.9 ± 65.9 mg at baseline and 1241.5 ± 53.3 mg at posttesting. For controls, dietary calcium intake was 1232.1 ± 70.28 mg at baseline and 1242.5 ± 65.7 mg at posttesting. These results are based on returned questionnaires from 74 of the 89 children. Values reported for our population are slightly higher than the national average of 1200 mg for children.(32)


Our aim was to study the effects of a high-intensity jumping program on hip and lumbar spine BMC in the growing skeleton. We report that 300 repetitions per week of jumping that produced ground reaction forces of eight times body weight resulted in significant improvements in femoral neck and lumbar spine BMC after 7 months.

Strengths of this study include randomization, pubertal status, use of a highly specific exercise program, and the measurement of peak ground reaction forces. First, the randomized controlled design of intact classrooms eliminated self-selection bias into the exercise group, assured equivalence between groups, and aided in minimizing the potential influence of hormones. Second, all children were classified as Tanner stage I (prepubescent), reducing the influence of sex hormones on bone. One girl advanced to Tanner stage III at posttesting and was excluded from the final analysis because the development of secondary sexual characteristics is hormonally controlled. Third, this is the first exercise intervention in children to use one specific exercise to increase bone mass at two clinically relevant fracture sites, the hip and lumbar spine. Other exercise interventions in children use a variety of different exercises, with varying intensities and durations, making it difficult to ascertain the specific exercise responsible for stimulating skeletal mineralization.(33–35) Last, the calculation of peak ground reaction forces in a subsample of children from the jumping and control groups allowed us to quantify the forces associated with jumping from a 61-cm-high box. Because ground reaction forces were not assessed at baseline, we had children from our control group perform 100 jump trials to determine peak ground reaction forces in children unfamiliar with the task of jumping. The resultant forces were similar between those children with experience in the task of jumping (jumping group) and those children with no experience in the task of jumping (control group). Because peak ground reaction forces were similar between groups (eight times body weight), the forces may have been similar across the intervention period.

An important study limitation is the inability to accurately detect changes in bone geometry caused by the two-dimensional nature of the DXA assessment. Both BMD and BA are imperfect variables, which only capture the height and width of the bone, without assessing the depth of the bone. However, BMC reflects changes in the true cross-sectional area of the skeletal region being examined.(31) This may, in part, explain the lack of uniform responses in area and BMD from training at the femoral neck and lumbar spine.

To date, a limited number of studies have examined the effect of exercise on growing bones, all reporting a positive bone response.(33–35) In an 8-month school-based jumping program, McKay and coworkers(33) found that children between the ages of 6.9 and 10.2 years who engaged in bone loading activities three times per week for 10–30 minutes had a 1.2% greater increase in femoral trochanteric BMD than controls. In this study, common games such as tag were altered to include hopping and bounding, in addition to 10 tuck jumps performed at each exercise session (estimated ground reaction forces between two and five times body weight). In this study a majority of the children were classified as Tanner stage I; however, some girls had advanced to stage two. Although some children changed Tanner stages, when controlling for growth factors (height and weight), significant differences remained at the trochanter. However, no group differences in BMD were reported at lumbar spine or femoral neck after controlling for changes in height and weight. In a 10-month nonrandomized trial, premenarcheal girls (Tanner stages I-III) engaged in impact activities (i.e., soccer, football, and skipping) for 30 minutes, three times/week.(34) Results indicated that exercisers had a 4.5, 4.1, and 11.3% greater increase in BMC, BA, and BMD, respectively, at the femoral neck compared with controls and a 5.5, 2.8, and 3.6% greater increase in BMC, BA, and BMD, respectively, at the lumbar spine compared with controls. However, after controlling for increases in height and weight, no differences were observed for femoral neck BMD or lumbar spine BMC. These findings suggest that growth and stage of sexual maturation may have played an influential role in the resultant bone response at both the hip and the lumbar spine, reducing the influence of exercise. In an 8-month trial by Bradney and coworkers,(35) prepubescent boys from two schools were randomly allocated to an exercise or control group. Exercisers engaged in moderate-intensity physical education classes for 30 minutes, three times per week. The exercise group had significantly higher BMC and BMD at the femoral midshaft compared with controls and significantly higher BMD at the lumbar spine (L2-L4). However, in a separate analysis of the third lumbar vertebra there were no group differences in BMC or area. Results from these longitudinal investigations suggest that a variety of exercises are capable of stimulating an osteogenic response in the growing skeleton. However, because of the varied training protocols in these reports, it is difficult to ascertain which exercise or if the combination of exercises was associated with a training response. In addition, the training protocols within each investigation produced varying skeletal responses, with some positive skeletal responses removed after controlling for changes in growth.(34,35) Data from our study show significant increases in bone mass at the femoral neck and lumbar spine from a highly specific jumping program after controlling for growth.

Reports in adult premenopausal women using jumping exercises as a means to stimulate osteogenesis also have yielded increases in bone mass at the hip and the spine.(36,37) Heinonen and coworkers(36) reported premenopausal women to have significant increases in femoral neck and lumbar spine BMD of 1.6% and 2.1%, respectively, resulting from 20 minutes of an aerobic jump/step program, three times per week, with peak ground reaction forces between 2.1 and 5.6 times body weight. These percentage increases translated to adjusted mean difference scores of 0.012 (95% CI, 0.003–0.020) at the femoral neck and 0.015 (95% CI, 0.005-0.025) at the lumbar spine. Additionally, Bassey and Ramsdale(37) found that 50 jumps per day performed on the floor increased trochanteric BMD by 3.4%, but not femoral neck or lumbar spine BMD. Peak ground reaction forces were reported to be approximately two times body weight. Based on these results and those from our investigation in children who performed jumps at eight times body weight, impact exercise stimulates osteogenesis.

Researchers have reported higher bone mass at the femoral neck in boys compared with girls(6,8,26–30); however, it is unclear if this difference is attributed to genetics, hormonal differences, weight-bearing physical activity, or a combination of these factors. To date, no studies have examined whether this relationship may be altered by exercise. Although we observed differences at baseline between genders at the femoral neck, there were no group by gender interactions at baseline or at the completion of the intervention for all bone variables. This indicates that the bone response was attributed to the high-intensity nature of the jumping exercises and was not influenced by gender.

As expected, height, weight, and body fat increased significantly within groups over the intervention period but were not significantly different between groups. Thus, jumping exercises stimulated an osteogenic response at the hip and lumbar spine without impeding increases in height, weight, or body fat, all of which are important for growth and development.(26–30) The children in our study were slightly above the national average for calcium intake.(32) However, we found no correlations between calcium intake and any of the bone variables. Data that examine the interactions between calcium intake, bone, and exercise are limited.(39) In an exercise intervention trial, infants between 6 and 18 months old were randomized into a 1-year activity program of bone loading exercises. Results indicated that children with low calcium intake had reduced BMC after performing the bone loading exercises; however, those with normal calcium intakes had had greater increases in bone mass than controls. Thus, it was postulated that children who participated in exercise during growth might lose bone if calcium is inadequate. In cross-sectional investigations, calcium intake has not been associated with physical activity. By contrast, physical activity has yielded greater gains in bone despite low calcium intake.(40–41) In the present study it is not known if the exercise intervention would have been as effective if reported dietary calcium intakes were lower. Further investigations should examine the interaction between calcium intake and exercise in the growing skeleton.

No major injuries occurred from the jumping activity. The nature of the jumping exercises may present concern for long-term health implications; however, there are no sporting activities that are free of injury and injury rates in youth sports are reported to be very low.(42) It is important to note that the majority of injuries that occur from high-impact loading activities occur when landings are performed incorrectly.(43) The children in our study were monitored carefully to ensure that the jumping exercises were performed safely and correctly. We believe that the intensity, frequency, and number of jumps performed in this study were safe for children of this age. There is no indication that overtraining occurred, because both exercise and control groups had similar gains in height, weight, and body fat. This is contrary to reports in children and adolescents who engage in intense exercise regimens that can result in a reduction in height, weight, and body fat.(38) Thus, we believe that the jumping exercises performed in this study will not lead to long-term health problems. However, we intend to follow these children to substantiate this claim.

Increased bone mass at the femoral neck and lumbar spine are powerful predictors of hip and spine fractures(1–2,44–45); thus, higher peak bone mass at these sites may reduce osteoporosis-related fractures. There is evidence that gains in bone mass during childhood will offset age-related bone loss. For example, in retired athletes the benefit of exercise is maintained into adulthood,(9,12-15) with greater maintenance observed in those athletes that commenced training before puberty.(12,15) Our study provides evidence that a simple jumping program offered in the prepubertal years may increase peak bone mass at two clinically relevant sites, the hip and lumbar spine. Long-term follow-up will provide evidence as to whether or not these gains are maintained over time and thus potentially reduce fracture risk in adulthood.(2)


We thank all of the parents and children who graciously volunteered to participate in this study and Mt. View Elementary school. This study is supported by National Institutes of Health grant RO1 AR45655–01, Division of National Institute of Arthritis and Musculo-Skeletal Diseases.