Effect of Starting Age of Physical Activity on Bone Mass in the Dominant Arm of Tennis and Squash Players

Methods

Participants

We recruited 105 currently ranked national-level female tennis and squash players for our study through the Finnish tennis and squash federations. The ethical committee for clinical investigation at our institute approved the investigational protocol, and we obtained informed consent from all participants. Ninety-seven players were right-handed (played with the right, dominant hand), and the remaining eight were left-handed. The mean age of players was 27.7 ±11.4 years (±SD). They were clinically healthy with no known diseases and were not receiving medications known to affect bone metabolism; none had previously had upper extremity fractures. The players' active training history had to be 5 years or more (mean, 10 ±6 years). The mean starting age of the playing career (that is, the age at which the athlete started to practice at least 2 sessions each week on a regular basis) was 16 ±9 years. They trained 4.4 times per week on average, and the average duration of each session was 80 minutes (range, 60 to 180 minutes). None of the participants performed or had performed activities, other than playing the racket sport, that affected only one extremity.

For the control group, we recruited 50 healthy Finnish women from local schools and work places. All but 2 of them were right-handed (that is, the right hand was dominant). The mean age of this group was 27.2 ±9.2 years. All participants in this group were also clinically healthy and had had no previous upper extremity fractures. Although some of them did participate in casual recreational sports (such as jogging, biking, skiing, swimming, and aerobics), none was involved in intense physical training or activities or work affecting the dominant or nondominant arm only.

Interview

The participants received a mailed questionnaire, which they completed independently at home. At the session during which anthropometric and strength measurements were obtained, one of three investigators (PK, HH, or MS) quickly reviewed the questionnaire responses with participants. This review determined whether the participants had understood and answered all questions. The three procedures (anthropometric measurements, strength measurements, and questionnaire review) were done in random order and always after bone measurements were obtained. The investigator was blinded to the bone measurement results.

The questionnaire included data on years of active playing, starting age of playing, number of training sessions per week, training intensity, average duration of each session, physical activities other than tennis or squash playing, injuries, medication, known diseases, diet, possible vitamin or mineral supplementation, consumption of alcohol, and use of cigarettes. We assessed the daily dietary calcium intake using a prospective 7-day questionnaire on consumed food, and we analyzed the results using Micro-nutrica software (Social Insurance Institution, Helsinki, Finland).

We also asked all participants about the age at onset of menses. We determined the menstrual status and divided the participants into three categories: 1) normal cycle of 23 to 35 days, with or without use of low-dose oral contraceptives, 2) any irregularity in menstrual pattern [such as short or long period, anovulatory cycles, short luteal phase, or oligomenorrhea], and 3) amenorrhea (no menstruation during the previous 6 months). We also asked the participants whether they had ever had disturbances in menstruation and the duration (in years) of such disturbances.

To test our hypothesis that the biological age at which the playing career was started was important for the development of the side-to-side difference in bone mass, we divided the players into six groups according to the starting age of playing relative to the age at menarche: more than 5 years before menarche, 3 to 5 years before menarche, 2 to 0 years before menarche, 1 to 5 years after menarche, 6 to 15 years after menarche, and more than 15 years after menarche. This division was based on the general knowledge of the pubertal and growth development of healthy Finnish and other white girls [11-16]: Puberty, once begun, is generally complete within 3 years; growth spurts and accelerated natural bone accumulation begin at the onset of Tanner stage 2, reach a peak at stages 3 to 4, and end at stage 5; menarche usually occurs during stage 4; and the longitudinal growth and natural bone accumulation rates markedly decrease soon after menarche, so that increases are only minimal in Tanner stage 5. Thus, the women in the six groups could be named as players who had started their playing careers at childhood [mean starting age, 7.4 ±1.4] years) prepuberty (10.1 ±1.2 years), puberty (12.0 ±1.4 years), postpuberty (15.2 ±2.4 years), early adulthood (24.0 ±3.0 years), and adulthood (33.7 ±3.8 years).

Anthropometric Measurements

We measured the height and weight of each participant. Using a measuring tape, we determined the circumference of upper extremities. We measured upper arm circumference just below the lateral part of the triceps brachii muscle and measured forearm circumference at the middle of the medial epicondyle of the humerus and the styloid process of the ulna.

Strength Measurements

We determined the maximal isometric strength of upper extremities using an arm flexion-extension dynamometer (Digitest, Inc., Muurame, Finland). We measured grip strength using a standard grip strength meter.

Bone Mineral Measurements

Using a Norland XR-26 DXA scanner (Norland, Inc., Fort Atkinson, Wisconsin), a technician determined bone mineral content (expressed in grams) from four sites in the upper extremity (proximal humerus, humeral shaft, radial shaft, and distal radius) and from the right calcaneus. The same experienced laboratory technician did all measurements. Her day-to-day coefficient of variation for repeated bone mineral content measurements of the same participants was low, ranging from 0.5% to 1.2% depending on the site measured [17, 18].

Statistical Analyses

We made intra-individual side-to-side comparisons using the matched, paired t-test. We used the Student nonpaired t-test to compare the continuous-type background variables, arm and calcaneus bone mineral content, and percentage of side-to-side differences among the players and controls. To compare the noncontinuous background variables of players and controls, we used the chi-square or Fisher exact test.

We tested the players' side-to-side bone mineral content differences across the six groups of players using analysis of variance, analysis of covariance, and a test for linear trend for adjusted group means. According to the previously noted hypothesis that puberty is critical in natural bone accumulation and that menarche is the first sign of cessation of bone development, the analysis of covariance was designed to include 5 (6 − 1) orthogonal or pairwise uncorrelated contrasts: 1) starting playing no later than at menarche (the first three groups) compared with starting thereafter (the remaining three groups); 2) starting no later than 3 years before menarche (the first two groups) compared with starting at menarche (the third group); 3) starting more than 5 years before menarche (the first group) compared with starting 3 to 5 years before menarche (the second group); 4) starting 1 to 5 years after menarche (the fourth group) compared with starting thereafter (the remaining two groups); and 5) starting 6 to 15 years after menarche (the fifth group) compared with starting more than 15 years after menarche (the sixth group).

In the analysis of covariance, we considered age, height, weight, body mass index, age at menarche, years of training, average number of training sessions per week, average duration of the training sessions, total training time (years of training times the average number of training sessions each week x the average duration of the training sessions), daily calcium intake, and the relative strength variables (elbow flexion, elbow extension, and grip) as potential confounders. Previous or present menstrual disturbances, smoking, use of alcohol, and special diets, all of which are potential confounders, were so uncommon among the players (Table 1) that these variables did not confound the results.

Our results are expressed as the mean ±SD. The 95% CIs are presented when appropriate. The given significance levels refer to two-tailed tests. We considered a P value less than 0.05 to be significant.

Results

Players Compared with Controls

No significant differences were seen between the background characteristics of the players and those of the control participants (Table 1). The dominant-to-nondominant side difference in upper arm and forearm circumference and muscle strength was significantly greater in players than in controls (Table 2). Compared with that of control participants, the dominant-to-nondominant side difference in bone mineral content was significantly greater in players in every measured site (players, 8.5% to 16.2%; controls, 3.2% to 4.6%) (Table 2). The difference was especially clear in the humerus.

When we compared the dominant arm bone mineral contents of the players and controls, the players' values were significantly greater in every measured site except the radial shaft (8.2% to 10.3%) (Table 2). The bone mineral content of the right calcaneus was also significantly greater in players (10.6%) than in controls. In nondominant (unloaded) arms, the differences between players and controls were small and not significant (range, −2.1% to 0.1%).

Players' Bone Results according to the Biological Starting Age of Playing

Correlation analysis showed that only age and height correlated significantly with both the independent or predictive variable (the biological age at which training was started) and dependent or outcome variable (side-to-side difference in bone mineral content); these two variables were therefore used as covariates in the analysis of covariance. This was confirmed by the fact that addition of any of the other potential confounders did not change the results adjusted for age and height.

At each anatomical site, the players' side-to-side bone differences were highly significant across the groups (analysis of variance: P < 0.001 to P = 0.005), the group means clearly decreasing with increasing biological age at which playing was started (Table 3 and Figure 1). For example, the difference was two to four times higher in the players who had started their playing careers before or at menarche (average difference, 10.5% to 23.5%) than in those who started more than 15 years after menarche (range, 2.4% to 9.6%). These great differences could also be verified by the fact that the CIs never overlapped among these groups (Figure 1).

Results of the trend test showed that findings at any site were not changed by adjustment for age and height (Table 3). In the analysis of covariance, this procedure did not alter findings in the proximal humerus or humeral shaft, whereas after this operation the group differences were no more significant for the radial shaft and distal radius (Table 3).

Of the five orthogonal contrasts of the analysis of covariance, the first contrast (players who started their playing careers no later than at menarche compared with those starting after menarche) was significant at every measured site except the distal radius. None of the remaining four contrasts was significant at any site.

Discussion

Menarche and completion of longitudinal growth are the first signs that bone mass development is stopping; 2 years after these developmental milestones, the expected increase in bone mass is only marginal [11-15]. For this reason, investigators have thought that the effects of the extrinsic bone mass determinants, such as calcium intake and physical activity, could also be strongest during this phase of rapid bone accumulation.

Our study provides evidence that physical activity during the pubescent years is crucial for maximizing bone mass (Table 3). Our findings did not ignore the potential for exercise-induced bone gain in adults but clearly showed that the benefit is about two times greater if women start their playing careers at or before menarche (humeral side-to-side difference, 17% to 24%) than after menarche (8% to 14%) (Figure 1).

Our study had several limitations. First, because our findings were derived from a selective group of healthy women who participated in a sport that strengthened the dominant arm, we cannot generalize the results to the general population. Second, although tennis and squash strokes seemed to cause considerable hypertrophy in the proximal humerus and distal radius, both of which are characteristic sites of osteoporotic fractures, our study did not show what type of activity would be the most beneficial in increasing bone mass in the other characteristic sites of these fractures (such as the spine, hip, knee, and ankle). Third, we do not know whether the additional bone mass in the playing arm is preserved throughout adulthood (with or without exercise). This question requires further follow-up of our players. Finally, an important undetermined issue is the minimum amount of exercise and minimum number of years of playing needed to achieve the observed increases and whether more exercise could further increase bone mass.

Our cross-sectional study suggested that tennis and squash activity started no later than puberty is maximally beneficial for mineralization of the bones of the playing arm, but longitudinal studies are needed.