Subjects and methods

Informed consent was obtained from the parents and from each of the 179 healthy adolescents (92 boys and 87 girls) enrolled in the first year of the Tii Junior High School in Shatin, Hong Kong. Children underwent physical examination by a fully trained paediatrician to exclude any condition known to affect bone metabolism. No pupil was excluded on this criterion, and none was taking medications known to alter bone or calcium metabolism, or both. The study was approved by the ethical committee for research on human subjects of the medical school of the Chinese University of Hong Kong.

Ninety four of the pupils enrolled were in the physical education major class (PE), and the other 85 were in the art major class (ARTS). Pupils had been assigned to the PE or ARTS classes according to their own and their parents’ choice and to school matching procedures, based on the past performance of each pupil in the major subject chosen.

LEVEL OF SPORTS PARTICIPATION

Each pupil was interviewed by one of the investigators to determine the type (ball games, water sports, track and field, and other) and the level (national/school team or recreational) of sport practised. In the PE group, 71.7% of the boys belonged to the national/school team, whereas none of the boys in the ARTS group did (p < 0.001).

DIETARY NUTRITION AND CALCIUM INTAKE

Each pupil was interviewed separately by a fully trained dietitian. A quantitative food frequency questionnaire,18modified and validated for the Hong Kong Chinese population,9 14 was used to assess individual dietary intake during the previous two months. Nutrient intake of the various food items was calculated from a comprehensive computerised food database compiled from previously published sources.19-22

ANTHROPOMETRIC MEASUREMENTS AND PUBERTAL STAGING

All anthropometric measurements were performed by one of two fully trained paediatricians, whose intratester and intertester variability was such that the results for each measurement were reproduced within 5% of the initial value in at least 80% of cases. Height without shoes was measured to the nearest 0.1 cm using a portable Harpenden Stadiometer (Holtain, Crosswell, UK).23 Weight in light vest and knickers was measured to the nearest 0.1 kg using a Seca digital physician’s scale (Seca Ltd, Bonn, Germany).24 25 Pubertal staging was done according to Tanner’s criteria.26 Menarcheal status was assessed by individual interview. Limb dominance was determined by asking the children which hand they wrote with and which leg they kicked a ball with.

BONE MASS MEASUREMENTS

The bone mineral content (BMC) and the width of both distal radii was measured using the Norland 2780 single beam photon absorptiometer (Norland Corporation, Fort Atkinson, Wisconsin, USA) (SPA). The BMD of the L2 to L4 lumbar vertebrae was measured using the Norland Dual-energy x ray absorptiometer (DEXA), Model XR-26 (Norland Corporation). These measurements were taken and the machines calibrated daily in accordance with the manufacturer’s instructions by one of two trained technicians.

The reproducibility in evaluating the radius in two successive scans without repositioning of the subjects were 2.09% and 2.06% respectively for BMC of the distal radius and bone width. Reproducibility of the DEXA scan results was within 1.4% for the L2 to L4 lumbar vertebrae.

PHYSICAL FITNESS ASSESSMENT

The following tests were performed in the order reported hereinafter according to the instructions of the American College of Sports Medicine (1995), the American Alliance for Health Physical Education, Recreation and Dance,27 or the manufacturers of the testing apparatus as appropriate. All tests were done after the children had performed stretching exercises for five minutes and cycled for a further five minutes on a commercially available cycle ergometer (Monark model 818E, Monark AB, Stockholm, Sweden) at a load setting of zero. The dominant limb was tested first, and the non-dominant limb was tested after a three minute rest (handgrip test) or a five minute (isokinetic test) rest. A three minute rest was allowed between each test. Children were verbally encouraged to produce maximal effort. The tests included (1) recovery index after a three minute step test28; (2) sit and reach test,29 using the Sit-and-Reach 5111 testing kit (Takei & Co Ltd, Tokyo, Japan); (3) one minute sit up test; (4) bilateral handgrip,27 using the Grip D 5101 dynamometer (Takei & Co Ltd, Tokyo, Japan); (5) vertical jump29; and (6) bilateral isokinetic quadriceps and hamstring peak flexion and extension torque test at 60º/second and 180º/second.30 A Cybex II⁺ dynamometer (Cybex, Lumex Inc Ronkonkoma, New York, USA) equipped with a HUMAC computer system (Computer Sports Medicine Inc, Totowa, New Jersey, USA) was used. The variables measured were peak torque (Nm), average work (joules), and average power (watts). Each test consisted of a trial phase of four submaximal and two maximal contractions at the given testing speed, a one minute rest, and an actual test phase of three continuous maximal contractions. Knee extension was tested first at 60º/second, and then at 180º/second, with a two minute rest between the two speeds. After a further two minute rest, knee flexion was tested starting at 60º/second and progressed to 180º/second after a two minute rest. The non-dominant limb was tested in the same way after a five minute rest.

For tests 2, 4, 5, and 6, three attempts were allowed, and the best result used for analysis.

PHYSICAL ACTIVITY ASSESSMENT

To assess physical activity over the past six months, each pupil was individually interviewed and information gathered through self report on: (a) type of major regular sports—ball games, track and field athletics, water sports, others; or (b) level of sports participation—PE group: national/school team members; PE group: recreational purpose; ARTS group.

STATISTICAL ANALYSIS

Descriptive statistics were calculated. The relations were studied between anthropometric measurements, pubertal staging, physical fitness, level of physical activities, and BMC of the distal radius and BMD of the L2 to L4 vertebrae. Univariate analyses by Student’st test, χ² test, and Pearson’s correlation coefficients were used to identify significant variables, which were then used for stepwise multiple regression analysis. The analyses were done separately for each sex to allow for the different measures of pubertal stages. Factors commonly predictive of BMC and BMD (for example, age, weight, height, and bone width) were given higher priorities to enter the regression model. Our goal was to identify other possible predicting factors. p values less than 0.05 were considered significant. SPSS for Windows version 6.0 was used for the statistical analyses.31

Results

ANTHROPOMETRY AND PUBERTY STAGING

Age, body weight, and height were not significantly different between the two groups. Of the boys in the PE group, 25% had their genitalia at Tanner stage 4, 15% had pubic hair at stage 4, and 43% had a testicular size greater than 12 ml (table 1). Among the girls in the PE group, 78% showed a stage of breast development of 4 or 5. Pubic hair had reached stage 4 or 5 in 32% of the girls in the PE group, and 63% of them were post-menarcheal. Children in the ARTS group showed similar findings (p > 0.05).

BONE MASS, CALCIUM INTAKE, PHYSICAL ACTIVITIES, PHYSICAL FITNESS, AND STRENGTH

There were no significant differences between the BMC of the dominant and non-dominant radii, and between dominant and non-dominant limb in handgrip and isokinetic testing (paired sample ttest). For subsequent analysis the dominant limb values were used.

Among the boys in the PE group, the mean (SE) BMC of the distal radius (0.631 (0.015) g/cm) was greater compared with boys in the ARTS group (0.573 (0.015) g/cm) (p = 0.009). Boys in the PE group also had a greater BMD of the L2 to L4 vertebrae (0.658 (0.019) g/cm²) v 0.587 (0.012 g/cm²) (p = 0.003). Calcium intake and bone width of the distal radius were not significantly different among the two groups (table 2).

Among the girls, there were significant differences between the PE and ARTS groups in all variables examined except calcium intake, BMC of the distal radius, and grip strength. Girls in the PE group had a greater mean (SE) BMD of their L2 to L4 vertebrae (0.792 (0.018) g/cm²) compared with those in the ARTS group (0.721 (0.015) g/cm²) (p = 0.002). The bone width of their distal radius was also larger (1.065 (0.015) cm v1.001 (0.014) cm) (p = 0.002) (table 2).

The mean (SE) BMC of the distal radius among boys (0.607 (0.011) g/cm²) was smaller compared with girls (0.643 (0.009) g/cm²) (p = 0.014). Boys also had a smaller BMD of the L2 to L4 vertebrae (0.628 (0.013) g/cm²) compared with girls (0.754 (0.012) g/cm²) (p < 0.001). This probably reflects the larger proportion of girls who were at a more advanced pubertal stage.

BONE MASS: UNIVARIATE ANALYSIS

Tables 3, 4, and 5 summarise the results of the univariate analysis of the relation between independent variables (calcium intake, anthropometric measurements, pubertal stage, bone width, physical fitness, physical activity) and BMC of the distal radius and BMD of the L2 to L4 vertebrae.

In boys, age, anthropometric variables, physical fitness variables, and pubertal staging were positively associated with the BMC of the distal radius and BMD of the L2 to L4 vertebrae. The length of L2 to L4, bone width of forearm, whether the pupils were in the PE or in the ARTS group, and most of all physical fitness tests were positively associated with BMC of the distal radius and BMD of the L2 to L4 vertebrae. Calcium intake and type of sport were not associated with BMC of the distal radius and with BMD of the L2 to L4 vertebrae.

The situation was similar for girls. However, BMC of the distal radius showed a negative correlation with calcium intake. Whether the girls were in the PE or ARTS groups just failed to show a significant effect on BMC of the distal radius. Calcium intake did not show a significant association with BMD of the L2 to L4 vertebrae, but it was significantly associated with the type of major subject class (PE or ARTS) in the girls (p = 0.002).

BONE MASS: STEPWISE REGRESSION ANALYSIS

The analysis identified the variables predictive of BMC of the distal radius and of BMD of the L2 to L4 vertebrae. Age, weight, height, and bone width were given high priority to enter the stepwise model, as they have been identified as significant predictors of BMC of the distal radius and BMD of the L2 to L4 vertebrae.8 9The other variables listed in the previous sections—that is, pubertal stage, group, level of sport, type of sport, physical fitness, and calcium intake—were all given equal chance to enter the model. Stepwise regression analysis showed that many of the factors, including inclusion in the PE or in the ARTS group, became non-significant. Table6 summarises the remaining significant predictive factors.

Among boys and girls the common positive predictors of BMC of the distal radius were age, body weight, and bone width of the distal radius. Flexibility in boys and pubic hair stage in girls were found to be additional significant factors.

Findings were similar for BMD of the L2 to L4 vertebrae. Age and body weight remained significant positive predictors. Vertical jump score and sit up test score, pubic hair stage in boys and girls, respectively were positively correlated with BMD of the L2 to L4 vertebrae. Bone width of the distal forearm did not predict BMD of the L2 to L4 vertebrae.

Discussion

Peak bone mass is attained in early adulthood and, after a period when it remains relatively constant, a slow progressive decline is observed.6 Actual bone mass and rate of decline are affected by multiple factors, the most important of which are genetic make up, age, sex, calcium intake, physical exercise, and other environmental lifestyle factors. Genetic influence could account for 70–80% of peak bone mass.8 9 Timing of the attainment of peak bone mass varies among different ethnic groups. For example, black people achieve a significantly higher bone mass and at an earlier age than white people do.8

Intervention studies have shown that in adult and older women11 and children13 sufficient calcium supplements can significantly increase bone mass even in children with calcium intake within recommended limits.13-15 Most researchers have found that programmed physical exercise and training, quantity of exercise, endurance exercise, and weight bearing exercise exert a significant positive effect on bone mass acquisition.7 10 12 14

Studies in children have shown the increase of bone mass with age, the correlation with calcium intake, body weight and height, and a significant difference between boys and girls and between different ethnic groups. However, detailed studies on the effects of physical exercise and of puberty on bone mass acquisition are relatively few,32-34 and their results are controversial. For example, a recent study from Canada showed that weight training in adolescent girls does not exert any significant influence on the increase of total body and lumbar spine BMD.

This study focused on adolescents and the various factors potentially affecting bone mass. These included pubertal stages, type and quantity of physical activity, physical fitness and strength, and calcium intake. Univariate analysis showed a significant positive correlation between BMC of the distal radius and BMD of the L2 to L4 vertebrae and age, weight, height, pubertal stage, bone width of distal radius, type of sport and physical fitness, and strength in both sexes. The effect of the different level of sports and frequency of sports participation were found to affect BMD of the lumbar spine more significantly than BMC of the distal radius. This is in keeping with other studies that have shown that BMD of the spine reflects more accurately the effect of physical exercise than BMC of the distal radius.35 The correlation with calcium intake was not significant, and the type of sport played affected neither BMC of the distal radius nor BMD of the L2 to L4 vertebrae.

When subjected to multiple regression analysis, it was confirmed that age, weight, and bone width are significant predictors of BMC of the distal radius and of BMD of the L2 to L4 vertebrae, at least in the study design adopted in our sample. Furthermore, BMC of the distal radius and BMD of the L2 to L4 vertebrae are also associated with some fitness variables in both sexes, and with pubertal stages. Pupils with higher pubertal stages therefore would be expected to have higher BMC of the distal radius or higher BMD of the lumbar spine.

In conclusion, BMC of the distal radius and BMD of the L2 to L4 vertebrae in boys were correlated with age, anthropometric variables, physical fitness, and pubertal stage. Calcium intake and type of sport practised did not exert a significant influence. In girls, while the results for BMD of the L2 to L4 vertebrae were similar to that found in boys, BMC of the distal radius showed a negative correlation with calcium intake. As in other ethnic groups, age and body weight were significant predictors of BMC of the distal radius. Physical fitness was a significant predictor of BMD of the L2 to L4 vertebrae, confirming the findings of other investigators that exercise has a significant effect on the lumbar spine.