The Side-to-Side Differences of Bone Mass at Proximal Femur in Female Rhythmic Sports Gymnasts



This cross-sectional study examined the side-to-side differences of the bone mineral density (BMD) at proximal femora in female rhythmic sports gymnasts (RSGs). The hypothesis on which the study is based is that gymnasts use a different leg in take-off (left leg) and in landing (right leg) and therefore differ in the loading for the left and right legs. The gymnasts made up two groups: the regular group, which consisted of 15 regular players who had trained for about 28 h/week, and the substitutes group, which consisted of 8 substitute players who had trained for about 12 h/week. The control group consisted of 10 nonathletic college women who had not participated in any kind of regular sports activity. BMD (g/cm2) was measured in three hip sites using the XR-26 dual-energy X-ray absorptiometer scanner. Muscle strength at knee extensors (EXT) and flexors (FLX) was examined using an isokinetic dynamometer (CYBEX6000), and the vertical ground reaction force was determined with a force platform during take-off and landing movements. In the regular players, the BMDs of the left leg were significantly higher than those of the right leg at the femoral necks, greater trochanters, and Ward's triangles (p < 0.01 ∼ 0.005). The side-to-side differences were 4.7 ∼ 9.6%. Regarding the strength parameters, the left side was greater than the right side significantly at EXT 60°/s (p < 0.01), although the overall side-to-side difference was small. In the substitutes, the BMDs at the three sites mentioned above were also higher in the left leg than the right, but the side-to-side difference was statistically significant only at Ward's triangles (9.3%, p < 0.05). The side-to-side difference of strength was not significant. In the controls, there were small left-to-right differences of the BMDs, ranging from −1.8 to 0.5%, which was significantly lower than in the regular players at each site. The overall average strength measurements were larger in the right leg than in the left except at the 120°/s. The side-to-side difference was statistically significant at EXT 30°/s and 60°/s (p < 0.05). The peak force was greater in take-off than in landing, and the unit time force during take-off was significantly greater than that during landing (p < 0.001). In conclusion, regarding the side-to-side difference of the BMD at proximal femora, our results demonstrate: that the left leg for take-off had higher measurements than the right leg for landing in both gymnasts' groups, which accounts for the vertical ground reaction force during take-off being greater than that during landing; that the difference in the regular players group was greater than that in the substitute group, which can be explained because the regular players practiced much more than the substitutes did; and that there was no difference in the control group.


THE IMPORTANCE OF PHYSICAL ACTIVITY in the development and maintenance of bone mineral density (BMD) is widely accepted. Cross-sectional studies have reported higher BMDs in athletes than in sedentary controls.1–3 Many follow-up studies have also been done. Most of these studies support the concept that exercise can slow down bone loss, maintain, and possibly even increase the bone density of humans; however, some studies have questioned this positive influence.4–7

Tennis and baseball, with their unilateral solicitations, allow a more precise examination of specific localized development.8–12 Using the tennis and baseball players and pairing study design with a side-to-side comparison, many confounding factors that are encountered in cross-sectional studies (such as genetic, hormonal, and nutritional factors) can be controlled.11 The above-mentioned unilateral activity had a clearly positive effect on the bone mass and muscle strength of the playing extremity.

The present study examined the differences of the BMD and muscle strength in both legs that have been used to take-off and land in leaping performed by rhythmic sports gymnastics (RSGs). The purpose was to assess the effect of long-term RSG on the BMD and muscle strength of lower extremities in female collegiate regular gymnasts and substitutes. The following questions were investigated by the present study:

  • Is there a positive association between long-term RSG training and the BMD and muscle strength based on the different functions of the legs?

  • How large are the side-to-side differences and are there relationships among the side-to-side differences and the athletic level of the regular players and substitutes?

  • Is there a lateral difference in the controls?



Subjects of the RSG groups were 15 regular players and 8 substitutes who met the following criteria: aged 18–21 years, without metabolic disorders and with no past lower extremity fractures, and performing RSG athletics—the left leg has been used to take-off, whereas the right leg has been used to land in leaping.

Various movement of RSG is attributed to jumping. These gymnasts train their left leg to jump up higher and farther and their right leg to land softly and safely to prevent injury. In terms of mechanical stress, at the moment of leaping, the left leg is under full stress to jump, whereas when landing, the right leg tries to reduce the stress caused by the shock of landing, letting the right toe touch the ground first.

The regular players were the participants in the National Championship and the World Championship. The substitutes were not eligible for those big games, but they practice with the regular players on a daily base. The substitutes practice less than the regular players due to the lack of time for special training and camp training for the big games.

The control group was comprised of 10 college students who were matched for gender, age, weight, and height to both the regular players and the substitutes groups. All of them had not performed regular sporting activities since their junior high school days, neither being a member of a sports club at school nor taking a sporting class in the community.

The subjects of this group were also clinically healthy and had no past lower extremity fractures. Before entering the study, all subjects were given an orientation and signed consent forms. They also completed questionnaires regarding their general health and menstrual and athletic histories.

Bone mineral measurements

The BMD (g/cm2) of the femoral neck (FN), greater trochanter (TR), and Ward's triangle (WT) of both side proximal femora were estimated using the Hip scan program of a Norland XR-26 dual-energy X-ray absorptiometry scanner (Norland Inc., Fort Atkinson, WI, U.S.A.). To minimize interobserver variation, all the scans were made by the same investigator, and his personal day-to-day coefficient of variation showed <1.0, 1.8, and 2.3% for the BMD at the FN, TR, and WT, respectively.

Muscle strength measurements

Muscle strength in foot-pounds was examined using an isokinetic dynamometer (CYBEX6000, Lumex, Inc., Ronkonkoma, NY, U.S.A.). The strength was measured in the knee extensors (EXT) and flexors (FLX) of the both legs. The measuring order should be reversed from one subject to the next, e.g., if one subject is measured first right and then left, the next subject should be measured left and then right. The subjects sat with a 105° hip angle and with the knee flexed at 90° with the lever of the dynamometer attached just above the ankle. The knee joint was aligned with the dynamometer's axis of rotation, and the angular movement of the knee joint was from 135–0°, that is, full knee extension and flexion. Peak torques were recorded at isokinetic velocities of 30, 60, and 120°/s. Three attempts were made at 30°/s and 60°/s, and eight attempts were made at 120°/s. All subjects were tested by the same examiner. The examiner and the subjects did not know the results of the bone mineral measurements. These measures of isokinetic strength were all closely correlated: r = 0.92 and 0.76 for the correlation of 30°/s with 60°/s and 120°/s, respectively, and 0.89 for the correlation of 60°/s with 120°/s.

Ground reaction force measurements

Force was recorded as the summed vertical force with the four cells of a Kistler force plate during take-off and landing in the RSG regular player group. The recording rate was 50 Hz, and representative data were the impulse, peak force, contact time, and impulse per contact time. The subjects were given as many practice leaps as needed to consistently take-off and land on the force platform, as they would on a normal gym floor.

Statistical analysis

The results in each group are indicated as mean ± SD. The side-to-side evaluation of the BMD and muscle strength and vertical ground reaction force was performed with the nonparametric Wilcoxon's signed-rank test. The training history variables of the two RSG groups were compared with the Wilcoxon's rank-sum test.

In the comparison of the three groups regarding the percentage of side-to-side differences, one-way analysis of variance was used. Statistical significance was set at p < 0.05.



The physical characteristics and training history of the groups are summarized in Table 1. There were no significant differences in age, height, or weight between the two RSG groups. Except for years of training, all values of training history were significantly higher in the regular player group than in the substitute group of RSGs (p < 0.001).

Table Table 1. Physical Characteristics of Subjects
original image

Menstrual status

Based on the definition of Drinkwater et al.,13 three gymnasts were oligmenorrheic and one was amenorrheic in the regular players group. All of the other subjects were eumenorrheic, having experienced 10–13 periods/year and having menstruated normally for at least 2 years. None of the subjects was taking the oral contraceptive pill.

Bone mineral

The average BMDs on proximal femora for each group are given in Table 2. Table 3 displays the percentage side-to-side differences among the three groups. In the RSG regular players, the BMDs were clearly and significantly higher at each site of the left limb (p < 0.01 ∼ 0.005). The side-to-side differences were from 4.7–9.6%, significantly larger than the controls at each site and significantly larger than the RSG substitutes in TR. In the RSG substitutes, all bone measurements were also higher in the left limb than in the right limb, and the side-to-side differences were statistically significant at WT only. On the contrary, in the controls, systematic side-to-side differences were extremely small: 0% and 0.5% at the FN and WT, respectively. Furthermore, in contrast with the two RSG groups, the differences were –1.8% at the TR.

Table Table 2. Left Versus Right Proximal Femur BMD and Isokinetic Strength of the Knee Flexors (FLX) and Extensors (EXT) Comparisons (Mean ± SD)
original image
Table Table 3. Comparison of Relative Side-to-Side Proximal Femur BMD Differences Among Three Groups (Mean ± SD)
original image

Muscle strength

Results for muscle strength are presented in Table 2. A statistically significant side-to-side difference was shown at EXT 60°/s in the RSG regular players (+8.4%, p < 0.01). In the RSG substitutes, although all differences between left and right were not significant, the largest difference (+3.4%) was similar to that in the RSG regular players, which was also higher in the left limb found at EXT 60°/s. In contrast, all peak torques were higher in the right limbs than in the left limbs of the control group, except at FLX 120°/s; the significant side-to-side differences existed at EXT 30°/s (−11.8%, p < 0.05) and 60°/s (−9.3%, p < 0.05).

Ground reaction force

Representative vertical force-time curves during take-off and landing conditions of the leap are shown in Fig. 1. The curve during take-off showed a single peak resulting from one intensive strike by the whole foot on take-off. However, the force curve during landing showed a double peak, which is considered a shock-absorbing technique. The curve of landing was characterized by longer contact times (325 ms) than that of take-off (204 ms). The impulse was also significantly greater in landing (p < 0.005). On the contrary, the peak forces were larger in take-off (2090.84 N) than in landing (1925.97 N) (p = 0.053). The unit time force (impulse per contact time) in take-off was significantly greater than in landing (p < 0.001) (Table 4). In addition, the peak force in take-off attained 4.3 times the body weight.

Table Table 4. Comparison of Vertical Ground Reaction Force and Contact Times During Take-off and Landing of the Leap Measured in RSG Regular Players
original image
Figure FIG. 1.

A curve of vertical ground reaction force components for take-off and landing conditions during the performance of a leap by a RSG regular player on the force platform.


This study has examined the side-to-side differences of BMD at proximal femora and muscle strength at knee EXT and FLX in female long-term RSG collegiate regular players and substitutes and in gender-, age-, weight-, and height-matched controls. The major findings of the study are: (1) regarding the RSG groups, the BMD values on the left side (take-off leg in leap) are higher than on the right side (landing leg in leap), the side-to-side difference averaged 7.0% in the regular players and 4.6% in the substitutes; in strength parameters, the percentage of side-to-side differences were small but statistically significant at EXT 60°/s in the regular players; (2) regarding the controls, there were few systematic left-to-right differences of the BMD, ranging from −1.8 to 0.5%, which were significantly lower than in the regular players group at three sites of the proximal femora; however, the overall average strength measurements showed larger in the right leg than in the left leg except the FLX 120°/s; the side-to-side differences were statistically significant at EXT 30 and 60°/s; and (3) regarding the force, the unit time force in take-off (left leg) was significantly greater than in landing (right leg).

This cross-sectional study was designed to determine the side-to-side differences in BMD of lower extremities that differ in the magnitude of mechanical loading on the basis of RSG special movements. This pairing design is often used to examine the effects of unilateral activity on BMD in comparing the playing and nonplaying arm, for example in tennis, baseball and squash. These results suggest that unilateral, usually intensive loading of the skeleton may affect the development of the BMD.8–12,14 However, no study has tested the influence of different mechanical loading on the bone of lower extremities with this study design. Our study found that significant ground reaction force (impulse per contact time) produced during take-off resulted in greater BMD compared with the counterpart's experiencing lower force imposed by landing in RSG athletics.

Biomechanical studies utilizing force platforms have shown that a high degree of symmetry in vertical ground reaction forces exists between the left and right limbs in runners and walkers.15 This result might explain why the side-to-side differences of the BMD in lower extremities were indeed small in the controls than the RSGs of the present study. On the contrary, in upper extremities of sedentary controls, the dominant-to-nondominant difference of BMD and muscle strength was significant.10–12,14 These findings suggest that the normal use of the dominant upper extremity causes noticeable changes in BMD and muscular strength.

The effects of exercise on bone mass differ depending on the elements: type, intensity, and frequency. Taaffe et al.16 and Fehling et al.17 have reported that gymnasts who chronically trained with weight-bearing loading had greater BMD at both axial and appendicular sites than the swimmers who exercised long-term without weight-bearing loading. These results suggest that the type of mechanical loading regimen plays an integral part in influencing BMD. Robinson et al.18) compared the bone mass of gymnasts and runners who exercised in the same weight-bearing environment but with different skeletal loading patterns. They found gymnasts had higher BMD than the runners and concluded that the higher impact force from the gymnastics training was thought to account for the differences. Furthermore, Kerr et al.19 examined the effect of a 1-year progressive resistance training program on BMD in postmenopausal women, using two training protocols which were different in the load applied and the number of repetitions. The results of this study suggested that the magnitude of loading is more important than the number of cycles or repetitions. In the present study, the impulse produced during take-off was lower than during landing, because the contact time for landing was longer than that for take-off. On the contrary, the peak force was larger in take-off (p = 0.053) and the unit time force (impulse per contact time) was significantly greater in take-off than in landing (p < 0.001) (Table 4). The BMD of the proximal femora were significantly higher in the take-off side than in the landing side. Thus, the peak force and unit time force were more effective in enhancing bone formation than loading time. This finding was also supported by the conclusion from the previous studies mentioned above.

The significant side-to-side differences of the BMD at the proximal femora were found in the RSG regular players. The RSG regular players were successful competitive athletes who performed special RSG programs 7 days/week, a minimum of 3 h/day. In contrast, the substitutes engaged in only basic training of RSG. Although the average training years were no different from the RSG regular players, the frequency, duration, and thereby the magnitude of skeletal loading were lower than in the RSG regular players. These differences produced the relative side-to-side differences of the BMD, which were significantly larger in the regular players than in the substitutes.

The side-to-side differences existing in the bone parameters were not consistent with strength measurements in the two RSG groups, except at EXT 60°/s. This inconsistency suggests that the influence of weight loading on BMD is very important, and this has been evidenced by studies of prolonged bed rest20 and observations during space flight.21 However, in the controls, the peak torques of EXT and FLX were higher in the right limb than in the left limb, and the side-to-side differences were large or reverse compared with the two RSG groups. This finding can be attributed to the fact that greater activity on the left leg increased muscle strength in RSG training. Therefore, the side-to-side differences of muscle strength in the RSG regular players and substitutes became small or reversed compared with the controls.

In the present study, four regular gymnasts with oligomenorrhea and amenorrhea show a tendency toward lower BMD than those with eumenorrhea, but their values were still high above those of the substitutes and the controls. This result is similar to Robinson et al.18 and Fehling et al.'s17 findings which reported that the menstrual disturbances in the gymnastic group did not seem to negatively affect the BMD at any site. Further, Slemenda and Johnston1 found that female ice skaters with oligomenorrhea and amenorrhea demonstrated only a minimal decrement in the lower body BMD (legs and pelvis) compared with normally menstruating skaters. They concluded that the unique loading characteristics (i.e., take-off and landing of jumps) of ice skating appeared to generate forces great enough to compensate for a deficient reproductive hormone profile in these athletes.

Our findings were also supported by the data of forces measured during jumps of RSG players. The peak force in take-off attained 4.3 times body weight, which was greater than in long distance runners, which was 2–3 times the body weight.15,22 In fact, low BMD caused by athletic amenorrhea was found mostly in the distance runners.18,23 That is, running is characterized by many more “cycles” of foot strike than gymnastics, and the lower load does not appear to stimulate bone formation.

In summary, this study examined the side-to-side differences of the BMD at the proximal femora in female RSGs. Our major findings are: (1) the left leg for take-off had higher measurements than the right leg for landing in both gymnasts groups, which accounts for the vertical ground reaction force during take-off being greater than that during landing; (2) the difference in the regular players group was greater than that in the substitute group, which can be explained because the regular players practiced much more than the substitutes did; and (3) there was no difference in the control group.