Differences in Bone Density, Body Composition, Physical Activity, and Diet Between Child Gymnasts and Untrained Children 7-8 Years of Age


  • The authors have no conflict of interest


Strategies that enhance the acquisition of bone mass may be protective against osteoporosis. BMD was compared in 20 artistic gymnasts (10 boys; 10 girls) and 20 untrained children ages 7-8 years. Higher regional values of BMD were observed in female gymnasts than untrained girls. If retained to adulthood, this higher BMD may protect skeletal integrity in later life.

Strategies that enhance the acquisition of bone mass in children may assist with the prevention of osteoporosis. This study explored the effects of regular high-impact and weight-bearing activity before the age of 7 years on total and regional bone mineral density (BMD). Twenty artistic gymnasts (10 boys and 10 girls) and 20 untrained children, 7-8 years of age, were recruited. The untrained children were matched to gymnasts by sex, height, weight, and age. Female gymnasts trained 8-10 h per week and had trained regularly for 3-4 years. Male gymnasts trained 4-6 h per week and had trained for 1-2 years. Measurements of bone mineral density were made using DXA for total body BMD (TBBMD); lumbar spine, both areal (aSBMD) and volumetric (vSBMD); total spine; pelvis; arms; and legs. Significant mean differences (8-10%) in aSBMD, vSBMD, arm BMD, and TBBMD were observed between female gymnasts and untrained girls (p < 0.05: aSBMD, vSBMD, and TBBMD body mass (BM); p < 0.01: arm BMD). A nonsignificant trend toward a higher TBBMD/BM and arm BMD was observed in male gymnasts compared with untrained boys. Trends toward a higher BMD within the pelvis, legs, and total spine were also observed in gymnasts. There were no differences in total and regional BMD between untrained boys and untrained girls. The results suggest that gymnastics training before the age of 7 years enhances the acquisition of bone mass at selected skeletal sites. The magnitude of this enhancement seems to be linked to the cumulative volume of such training. If retained during adolescence and young adulthood, a surfeit of bone acquired through high-impact and weight-bearing activity in early childhood may protect skeletal integrity in later life.


STRATEGIES THAT INCREASE the acquisition of bone mass during childhood may reduce the risk of osteoporosis in adulthood.(1, 2) Physical activity involving high-impact or weight-bearing movements provides an osteogenic stimulus that may enhance bone mass at any age.(3, 4) However, there is evidence to suggest that the capacity of bone to adapt its mass to such activity is greatest before puberty.(1, 2) Therefore, the practice of high-impact or weight-bearing activities in early childhood may lead to substantial gains of bone mass that, if retained, could help to protect skeletal integrity in later life.

One form of physical activity that has been shown to be particularly effective for the acquisition of bone mass in children is gymnastics training. Cross-sectional studies have shown that girls between 9 and 13 years of age, who have trained regularly in gymnastics for 2 or more years, display a higher bone mineral density (BMD) at specific skeletal sites than their untrained counterparts.(5–8) Two of these studies have also followed their subjects longitudinally and have confirmed that over a 12-month period, between the ages of 9 and 12 years, gymnasts accrue a greater quantity of bone than their untrained counterparts.(6, 7) However, an important point is that an interpretation of these findings is complicated by the differential rates of bone growth and mineral acquisition that children within such a broad age range can display. Furthermore, it is likely that some children within this age range would be close to, or have started, puberty, when a surge in the circulating levels of sex hormones exerts a dominant influence over bone growth and mineral accrual.(9, 10) This influence would complicate a partitioning of the effects of physical activity on the acquisition of bone mass in children. It is also noteworthy that these previous studies have investigated the effects of regular and specific physical activity on the BMD of girls only.

The purpose of this study was to compare total body and regional BMD in 7- to 8-year-old male and female artistic gymnasts with nonspecifically trained children matched for height, mass, and age. Simultaneous assessments of body composition, habitual physical activity, and diet were made in all participants. It was anticipated that the nature of training undertaken by children selected for an elite pathway in gymnastics would separate them from their untrained counterparts and therefore facilitate an exploration of the effects of regular high-impact and weight-bearing activity on the development of bone mass during early childhood. The originalities of this research relative to previous comparative investigations of BMD in child gymnasts and controls relate to the younger age and narrower age range of the participants and the study of boys as well as girls.



Forty children (20 boys and 20 girls), between 7.1 and 8.5 years of age, were recruited for this study. Subject numbers were calculated in accordance with the variation of bone density in untrained 7- to 8-year-old children measured within our laboratories, accounting for the precision of the apparatus as well as mean differences in BMD between trained and untrained prepubescent children from previous studies.(5, 6, 8) Sample size calculations were based on a statistical power of 80% and a significance level of 5%. One-half of these children (10 boys and 10 girls) were trained artistic gymnasts who had been selected for an elite pathway in this sport to represent Leeds City. The girls trained for performance in four apparatus: floor, vault, beam, and asymmetric bars, and the boys trained on six pieces: floor, vault, pommel horse, rings, parallel bar, and high bar, as stipulated by the International Gymnastics Federation. The remaining 10 boys and 10 girls, who acted as controls, were untrained and were recruited from primary schools within a region of Bradford, West Yorkshire, which is considered to represent an “average” socioeconomic status. The parents of all controls and 16 of the gymnasts described their ethnic origin as white. The remaining children (three male and one female gymnasts) reported as black African or Caribbean. Although there is evidence to suggest that racial differences in bone density may exist in children,(11, 12) the bone density measurements of the four Afro-Caribbean gymnasts were similar to those of their white counterparts, and we therefore included data from all children. Each gymnast was matched with a control of the same sex in accordance with height and as closely as possible with body mass and age. All gymnasts completed specific “conditioning work” within each training session, which focused on exercises to improve strength, local muscular endurance, and flexibility. The girls also performed additional work in apparatus. The girls trained for 8–10 h per week and had been training for 3–4 years, whereas the boys trained for 4–6 h per week and had been training for 1–2 years.

All participants were volunteers, and written informed consent was obtained from each child and a parent. The investigators met with all parents and children to explain the study procedures and to distribute associated information sheets 2 weeks before obtaining written consent. The parents of all children were asked to inform the child's general practitioner of their intended participation in the study, and only those children who were apparently healthy were included. No participant had a history of disease or adherence to medication that can influence bone metabolism. Ethical approval for the study was obtained through the Local Research Ethics Committee of Leeds Teaching Hospitals Trust, West Yorkshire.

Bone density and body composition

Areal BMD of the lumbar spine L2-L4 (aSBMD) and total body BMD (TBBMD) was measured using a DXA fan beam absorptiometer (GE/Lunar Prodigy, Madison, WI, USA). Both scans were performed in the thin (pediatric) scan mode. In this mode, the absorptiometer performs total body scans in 3–4 minutes and lumbar spine in less than 1 minute, thus reducing the problem of movement of children during scanning. From the total body scan, body composition measurements of lean tissue mass (LTM), fat mass (FM), %FM, total body bone mineral content (TBBMC), and regional BMD of the arms, legs, pelvis, and total spine were obtained. Scan analysis was made using version 5.0 software. BMD in vivo precision (%CV) for lumbar spine L2-L4 = 1.9% and total body = 0.8% were determined by measuring 10 children twice on the same day with repositioning between the measurements. Before the scan, the height (cm) and mass (kg) of each subject was recorded to the nearest 1 mm and 100 g, respectively.

Because DXA expresses BMC in relation to projected bone area (PBA), it provides a two-dimensional model that fails to account for differences in bone geometry. This model may not therefore provide an accurate measure of the amount of material within a bone. An estimate of bone volume, commonly referred to as bone mineral apparent density (BMAD), has frequently been used to facilitate a comparison of bone mass between growing children with different sized bones.(5, 7, 8) BMAD may be calculated by dividing BMC by PBA1.5(13) and was applied to the measurements made at the lumbar spine of our subjects. We also expressed BMD as a simple ratio function to mass, because in accordance with bone loading theories,(3, 14) an increase in body mass with growth operates in conjunction with weight-bearing activity to assist the acquisition of bone mass.

A standardized score (Z-score) of areal BMD at each measurement site was calculated for each gymnast by subtracting the observed value of BMD from the mean value for controls of the same sex and height and expressing the difference relative to the SD of the mean value for controls. Group means and SDs of Z-scores for each measurement site were then calculated.

Physical activity assessment

The habitual physical activity level of each child was estimated from a 7-day diary that was modified from a method described by Ainsworth et al.(15) This diary was structured around the school day for weekdays to facilitate recall and completion. Each child and a parent were given verbal and written instructions for completing the diary. This involved recording the duration, to the nearest 15 minutes, of activities that are categorized based on their estimated energy demands. Each activity category is designated a “metabolic equivalent” (MET) value, where 1 MET equates to resting metabolic rate (RMR). The MET value for an activity is therefore a multiple of RMR. The categories included the following activities in which the children regularly participated: 1 MET—sleeping, lying relaxed, watching television; 3 METS—leisurely walking, various indoor play activities; 5 METS—outdoor play, including ball games and bicycling; 7 METS—gymnastics training and other vigorous sports. The remaining time spent each day performing sedentary activities (1.5 METS) was calculated by subtracting the number of hours spent performing the categorized activities from 24. The RMR of each child was calculated using the technique of Schofield.(16) Mean daily energy balance (EB) was calculated for each child by subtracting dietary energy intake (EI) from estimated energy expenditure (EE) recorded over a 7-day period.

An estimate of the volume of high-impact activity undertaken for each child was obtained through the calculation of a “weight-bearing score” that was adapted from a procedure reported by Verheul et al.(17) This procedure involves the designation of a numerical value to a variety of activities that require the movement of body mass against the force caused by gravity. Each value is a multiple of body mass and equates to the magnitude of the vertical ground reaction force (GRF) produced while practicing the activity as determined using a force plate. For example, walking, running, and jumping have been shown to generate GRFs that equate to approximately 1.1, 2.5, and 6.0 times body mass, respectively. The vertical GRFs produced during common gymnastic exercises have been reported to equate to between 12 and 14 times body mass.(18) On the premise that it is only those activities that generate GRFs in excess of 2.5 times body mass that are likely to provide a significant osteogenic stimulus,(17, 18) activities producing GRFs below 2.5 were not included within the calculations of the weight-bearing score. Thus, the range of estimated GRFs extended from 2.5 times body mass (running) to 14 (gymnastic maneuvers such as vaulting, somersaults, and landing from high bars). For each child, the weight-bearing score was calculated by multiplying vertical GRF by the child's body mass and the duration that they practiced a high-impact activity for a period of 7 consecutive days.

Diet analysis

An estimate of the habitual dietary intake of each child was made from a diet diary that was completed by the child, with parental assistance, over the same 7-day period as their physical activity diary. The child and parent were asked to record all food and beverages consumed and instructed how to note quantities of foods and drinks consumed, “portion” sizes, different brand names, and plate waste. The importance of an adherence to the normal diet was emphasized. Food models and photographs of foods were used to assist the understanding of children and the accuracy with which the diaries were completed. Data from the diaries were analyzed using a nutrition database program (CompEat version 5.0; Carlson-Bengsten Consultants Limited, Grantham, Lincs, U.K). The data base of this software program contains a comprehensive range of fortified foods (e.g., breakfast cereals and fruit-flavored drinks) available in the UK and also permits the entry of food supplements to enable a complete diet analysis. This program differentiates the intake of nutrients occurring naturally within foods from those ingested as supplements. The age, sex, height, and mass of each child were entered into the software program to select Reference Nutrient Intake (RNI) values for the child. The RNI for a nutrient is defined by the UK Committee on the Medical Aspects of Food and Nutrition Policy (COMA) as the ingested quantity that covers the metabolic requirements of 97.5% of a selected population.(19) Particular attention was paid to the intake of energy, protein, and calcium, which have been identified as nutrients that play a particularly important role in the acquisition of bone mass in growing children.(1, 20)

Statistical analyses

Data sets were analyzed for normality of distribution and homogeneity of variance. These analyses confirmed the suitability of parametric tests to compare the mean values of data sets between groups. A factorial two-way ANOVA (single observation on separate groups: gymnasts vs. controls; male vs. female) was used to compare group mean values of age and anthropometric variables, BMD measurements, physical activity, and nutritional parameters. After the calculation of significant F values, a Tukey test was performed to compare specific mean differences. The magnitude of difference in selected (nonsignificant) BMD values between subject groups was also evaluated through the calculation of effect size (d): that is, difference between group means/pooled SD.(21) Because effect size expresses a group mean difference relative to the variance of a specific population, it enables an evaluation of comparative differences of a variable made in different studies investigating sample groups of different sizes. Small, moderate, and large effect sizes equate to values in the order of 0.2, 0.5, and 0.8, respectively.


Physical characteristics

Selected physical characteristics of the gymnasts and controls are shown in Table 1. The SD scores (SDSs) for height and mass are shown in Fig. 1. These SDSs suggested that both male and female gymnasts were smaller and lighter than a reference population of healthy children of the same age.(22) However, the heights and weights of untrained boys and girls (controls) concurred with the reference values. Because gymnasts were generally smaller for their age than untrained controls, the mean age of gymnasts tended to be higher than that of controls. Nevertheless, there was no statistically significant difference in age between the four groups. By design, the means and SDs of height and mass of children within the four groups were similar. For either sex, lean tissue mass (LTM) tended to be higher in gymnasts than untrained children: 5.2% (0.9 kg) and 6.3% (1.3 kg) for girls and boys, respectively, and 5.7% (1.0 kg) higher in untrained boys than untrained girls; however, these differences were not significant. Gymnasts also exhibited a lower fat mass (FM) than untrained children of the same sex: 45% (2.8 kg) and 43% (2.3 kg) deficit for male and female gymnasts, respectively (p < 0.01, both comparisons). Untrained boys had a 21% lower FM than untrained girls (p < 0.05).

Table Table 1.. Bone Mineral Density Measurements of Male and Female Gymnasts and Untrained Boys and Girls (Controls)
original image
Figure FIG. 1..

Height, weight, and regional areal BMD of male and female child gymnasts expressed as standardized (Z) scores. Height and weight are expressed relative to chronological age and sex using data from a reference population of healthy children.(22) BMD is expressed relative to mean values for untrained (control) children of the same sex and height. The white bar denotes females and the shaded bar denotes males. Both groups of gymnasts were small and light for their age, yet displayed trends toward a higher BMD than the controls. Female gymnasts displayed a significantly higher BMD than untrained girls within the lumbar spine (p < 0.05) and arms (p < 0.01). *p < 0.05; **p < 0.01, gymnasts vs. controls of the same sex.

Bone density

The BMD measurements of the four groups of children are summarized in Table 1. Figure 1 illustrates regional values of areal BMD within gymnasts, standardized as Z-scores. Significant mean differences (in the order of 9%) were found between female gymnasts and untrained girls (controls) at the lumbar spine, aSBMD (p < 0.05) and vSBMD (p < 0.05), and arms (p < 0.01). Compared with untrained girls, female gymnasts had a higher total body BMD when expressed as a ratio to body mass (i.e., TBBMD body mass (BM); p < 0.05), which was also in the order of 9%. Female gymnasts also displayed a higher aSBMD and vSBMD than either male gymnasts (p < 0.05) or untrained boys (p < 0.05). Differences or variations of BMD between the four groups were attributable to differences in BMC rather than to PBA because the latter was almost identical at each measurement site between the four groups. A nonsignificant trend toward a higher TBBMD/BM and arm BMD was observed in male gymnasts relative to untrained boys (effect size d = 0.55 and 0.66, respectively). Although TBBMD/BM was similar in male and female gymnasts, the untrained boys tended to have a higher TBBMD/BM than their female counterparts. Similar values for aSBMD and vSBMD were found within the two groups of boys and untrained girls. Trends toward a higher BMD in gymnasts relative to untrained children were also observed within the pelvis (d = 0.50 and 0.49; girls and boys, respectively), legs (d = 0.58 and 0.14, respectively), and total spine (d = 0.82 and 0.25, respectively). These effect size calculations show a greater magnitude of effect within girls than boys. The Z-scores for areal BMD at different skeletal sites within the female gymnasts (Fig. 1), which varied from 0.68 (total body) to 1.55 (arms) above control values, illustrated their apparent surfeit of bone. This figure also highlights the small stature and low body mass of both groups of gymnasts relative to the normal values for boys and girls of the same age.

Physical activity

A summary of the physical activity patterns of the four groups of children are given in Table 2. Analysis of physical activity diaries suggested that compared with untrained children, gymnasts undertook a significantly greater volume of physical activity that demanded an energy expenditure in the order of 7 METS (p < 0.001). This was evidenced by estimates of the total time that they engaged in such activities over 7 days of recording. Furthermore, the number of hours that female gymnasts engaged in such activity was in the order of 45% greater than that of male gymnasts (p < 0.01). The volume of physical activity undertaken within the 7 MET category tended to be greater in untrained girls than untrained boys but was not significantly different. Estimated energy expenditure reflected the amount of time spent undertaking vigorous activities by gymnasts and controls and was 21% and 19% higher in gymnasts than controls for males and females, respectively (p < 0.01, both sexes). Gymnasts undertook a significantly higher volume of high-impact activity than controls, indicated by differences in the calculated “weight-bearing score” (WBS), which was eight and five times higher in gymnasts than controls for females and males, respectively (p < 0.001, both sexes). The higher volume of training in female relative to male gymnasts was exemplified through the respective calculations of WBS, which was almost twice as high in the former (p < 0.01). Although the WBS for individual training sessions was found to be comparable between male and female gymnasts, the females undertook twice as many training sessions over a week. The calculated WBS did not differ in untrained boys and girls, and none of these children participated regularly in activities that generate ground reaction forces exceeding six times body mass. The higher WBS within gymnasts reflected their gymnastic training rather than other habitual activities. Thus, a deduction of the WBS of this training from the gymnasts' total WBS produced similar scores to those of untrained children.

Table Table 2.. Physical Activity and Dietary Patterns of Male and Female Gymnasts and Untrained Boys and Girls (Controls)
original image

Diet and energy balance

There were no differences in energy, protein, or calcium intake between any of the four groups. All children reported intakes of protein and calcium that exceeded the UK RNI for children between 5 and 8 years of age (28–30 g/d and 550 mg/d, respectively). In the case of protein, most children seemed to be ingesting an excess of 150–200% of their estimated requirement, whereas reported calcium intake was 120–150% above the UK RNI. No overt deficiency of any other nutrient was recorded within any of these children. Although none were taking calcium supplements, many regularly ingested foods that were fortified with vitamins and minerals that included calcium. Nevertheless, the primary source of calcium within the diets of these children was dairy products. Estimates of energy balance suggested that most children achieved an equilibrium between energy intake and energy expenditure; however, an apparent energy deficit of between 1000 and 1200 kJ was calculated in two of the female gymnasts and one male gymnast.


The main finding of this study was that the BMD of female artistic gymnasts, 7–8 years of age, with a 3- to 4-year history of regular training, was significantly higher within the arms and lumbar spine than that of untrained girls or boys of a comparable age who were matched for height and mass. These female gymnasts also displayed a higher lumbar spine BMD than a similarly matched group of male gymnasts with a 1- to 2-year history of training. Total body BMD was higher in female gymnasts than untrained girls when expressed relative to body mass, and there were accompanying trends toward a higher BMD within the pelvis, legs, and total spine.

The cumulative effects of 1–3 years of gymnastics training on bone mass in children as young as 7 years has not been explored previously. Furthermore, there have been no previous studies that have compared the BMD of young male gymnasts with untrained boys and girls or with female gymnasts matched for age, height, and mass. To date, related research has tended to focus on older girls (9–13 years, with a common mean age of 10 years), who have usually trained for between 2 and 6 years.(5–8) Our comparative observations of BMD in female gymnasts and untrained girls followed a similar trend to the findings of the aforementioned studies. The 10-year-old female gymnasts within these previous studies had an 8–12% higher BMD of the lumbar spine and a 3–5% higher total body BMD than their untrained counterparts, which concurred with the respective 9% and 3% differences that we measured between our younger female gymnasts and controls. Differences in arm BMD between gymnasts and controls has only been explored in a single previous study,(6) which reported an 11% higher value in gymnasts. This compared with the 9% difference in our study.

Despite a trend toward a higher BMD within the arms, pelvis, and total body, the BMD of our male gymnasts did not differ from that of untrained boys at any measurement site. A possible explanation is the duration that these gymnasts had undertaken regular and structured training for their discipline (13–22 months). This compared with a 38- to 46-month time period of organized training for the female gymnasts. At the time their BMD was measured, the male gymnasts were still following a preparatory phase of training termed “conditioning,” which emphasizes the enhancement of muscle strength and endurance, together with the promotion of flexibility within tendons and ligaments. They had yet to progress to the higher volume of training performed by the female gymnasts, which incorporated an additional focus on apparatus work. In contrast to the male gymnasts, the female gymnasts had already completed 2 years of preparatory “conditioning” training and had been performing apparatus work for at least 1 year. Altogether, these female gymnasts were undertaking nearly twice the volume of training as their male counterparts, which was scheduled into twice the number of weekly training sessions. These comparative observations of retrospective and current training practices between our male and female gymnasts might suggest that the cumulative volume of high-impact gymnastics training had exerted a significant influence on their BMD.

In addition to training history, the magnitude of bone loading in training and its rate of application are important determinants of skeletal adaptation.(3, 14) The frequency and volume of loading are considered to play a lessor role. Despite the similarity of the WBS calculated for each training session undertaken by our male and female gymnasts, it is noteworthy that this score considers only one type of osteogenic stimulus encountered during gymnastics activity: that of vertical GRF associated with landing on the lower limbs. Although such force exerts mechanical strain within the lower limbs, it loads the spine and upper body to a much lesser degree.(18, 23) Furthermore, the WBS does not quantify the magnitude of joint reaction force (JRF) generated by muscle pull during the movement of body mass against external resistance, which may be equally osteogenic.(3, 14) It is conceivable that the apparatus work practiced by the female gymnasts generated higher JRFs than that produced during “conditioning” exercise (practiced by both groups of gymnasts), and that such JRFs assisted the acquisition of bone mass in these girls. This concept could also explain the higher BMD observed at the lumbar spine of female relative to male gymnasts. An exercise-induced enhancement of BMD at the spine depends on the practice of high-impact activities applied to the upper body and/or weight-bearing activities that involve the recruitment of trunk and arm musculature to stabilize the spine.(3, 14) The apparatus training undertaken by the female gymnasts clearly involved such activities.

A major strength of this study was the deliberate recruitment of children within a narrow age range (7.4–8.5 years for gymnasts and 7.1–8.0 years for controls) who were below the expected age of puberty. The purpose of adhering to these selection criteria was to limit variance within the physical and metabolic characteristics that are known to influence the acquisition of bone mass in children. The primary determinant of BMD in prepubescent children is bone age, which is associated with chronological age, height, and mass.(9, 20) In this study, for either sex, we matched controls to gymnasts by height and then as closely as possible by mass and age. The tendency for gymnasts to be older than controls suggested a slight growth retardation in the former, which concurs with the findings of other studies.(6, 24) Previous studies that have compared the skeletal characteristics of child gymnasts and controls have focused on children within a 4- to 5-year range, which has resulted in a greater diversity of physical characteristics than we observed.(5–8) Furthermore, the selected age range for these previous studies has included children between 11 and 13 years of age, some of whom have started puberty, as evidenced by a progression of Tanner stage. Although we did not screen our subjects for pubertal development, it did not seem that any had entered a pubertal growth spurt, and none had a stature in excess of 0.85 SD above the age-matched mean value; thus, it is unlikely that sex hormones were exerting a significant effect on their bone development.

The magnitude of BMD in children has been shown to be positively associated with muscle mass.(25, 26) In this study, this trend was apparent when comparing gymnasts with untrained children of the same sex; however, despite a trend toward a higher LTM, male gymnasts displayed a lower BMD at most skeletal sites than their female counterparts. The differences in body composition that we observed between female gymnasts and untrained girls were of a similar pattern to the differences reported in previous studies, but of a lesser magnitude. The 10-year-old female gymnasts studied by Nickols-Richardson et al.(7, 8) and by Bass et al.(6) had a 10–12% higher LTM and a 30–60% lower percentage FM than untrained girls. This compared with a 5% higher LTM and a 43% lower FM in our 7- to 8-year-old female gymnasts and controls.

The influence of race on the BMD of prepubescent children has received little research attention, despite the documented difference in the BMD of adults of different racial origin.(27, 28) Although there is evidence to suggest that black children have a higher BMD in selected regions than white children,(11, 12) the black gymnasts in our study did not display a higher BMD at any of the measurement sites. However, we acknowledge the potential benefits of having been able to recruit an equal number of black controls. Nevertheless, the racial origin of a number of the children in this study was uncertain, and the challenge of isolating the effects of race from the many genetic and environmental influences on the development of bone mass in childhood should be recognized.

Nutrition plays a permissive role in skeletal growth and the acquisition of bone mass in children; that is, nutrient deficiencies may impair these processes, whereas excesses above a certain threshold do not enhance these processes beyond an inherent ceiling.(1, 20) Energy balance and the intake of calcium and protein have been highlighted as pertinent nutritional determinants of bone development.(1, 20) Because none of our child gymnasts or controls displayed nutrient deficiencies, it is unlikely that the observed variations of BMD could be explained by dietary factors. The observation of a calcium and protein consumption in excess of the UK RNI concurred with the findings of a recent UK dietary survey of children's eating habits across a variety of socioeconomic classes.(29) Despite the apparently low energy intakes of three of the gymnasts, the stature and body mass of these children fell within 0.7 SD of the age- and sex-matched mean values. Although changes in body mass over the period of collection of dietary data were not documented, the possibility of an alteration of habitual diet or error in the self-reporting of dietary intake by children or their parents might be considered.

There has been debate regarding the most accurate and meaningful mode of expression of BMC measured using DXA, particularly in growing children. In many comparative studies of bone mass in children using DXA, BMC has been expressed relative to estimated bone volume (i.e., PBA1.5) and referred to as bone mineral apparent density (BMAD).(5, 7, 8) Within these previous studies, greater differences in the estimated bone mass of the lumbar spine between female gymnasts and their untrained counterparts have been observed when BMC has been expressed as BMAD rather than as BMD. Nevertheless, the children within these studies have displayed a wide age range (4–5 years),(5–8) and groups with different physical activity patterns have sometimes been matched in accordance with chronological age rather than by height or estimated bone age.(5) In this study, the recruitment of children within a narrow age range, together with the matching of controls to gymnasts by height and mass, seemed to limit the requirement for an adjustment of measures of BMD at the lumbar spine for estimated bone volume. The magnitude of the difference in lumbar spine BMD and BMAD between female gymnasts and either male gymnasts, untrained boys, or untrained girls was observed to be similar. Furthermore, group mean values of BMC and PBA between all four groups were comparable.

Owing to the great skill and physical and mental demands of gymnastics, it would be unrealistic to prescribe this activity to the majority of children as a potential prophylactic to osteoporosis in adulthood. However, it is possible to devise physical activity programs that include modified gymnastic exercises, but which are independent of specific skill. Although genetics, particularly the inherent capacity to build bone in response to regular and specific physical activity, could account for a proportion of the apparent surfeit of bone in child gymnasts, the efficacy of regular high-impact and/or weight-bearing activity for the promotion of bone mineral acquisition in prepubescent children has been demonstrated in some recent randomized controlled trials.(23, 30–32)

In summary, the findings of this study suggest that bone responds to the osteogenic stimulus of gymnastics training in children before the age of 7 years. Differences in the magnitude and distribution of BMD between male and female gymnasts seemed to reflect their past and present patterns of physical activity, particularly the cumulative volume of high-impact and weight-bearing activities. To protect skeletal integrity in later life, it is essential that a surfeit of bone accrued through physical training in childhood is retained to adulthood. Further research is required to ascertain whether this surfeit remains if the osteogenic stimulus is removed through a discontinuation of the activity.


We thank the child participants and their parents for their enthusiasm and patience, and Mark Gannon and the head teachers of the primary schools for their assistance with the recruitment of control children for this study.