Introduction

Osteoporosis is widely recognized as a major health problem in aging men and women. In Hong Kong, the prevalence of osteoporosis among Chinese women and men aged 50 years old or above were 37% and 7% at the spine and 16% and 6% at total hip, respectively [1]. The risk of osteoporosis is determined by two main factors: the peak bone mass in adulthood and the rate of bone loss with aging. Much research has been conducted on bone loss in the elderly; however, little is known about bone accretion in adolescents.

During childhood and adolescence, bone mineral density (BMD) increases until peak bone mass is reached. Those who achieve a higher peak bone mass are less at risk of having an osteoporotic fracture later in life [2]. To prevent future osteoporosis, intervention carried out in the adolescent years is the most effective approach as individuals gain much of their peak bone mass by the end of adolescence [3].

The site-specific relationship between muscle strength and bone mass previously has been demonstrated in grip strength and forearm bone density in adults and elderly [4, 5]. Moreover, some studies concluded that the relationship between grip strength and bone mass is not only site specific but also systemic [6–9]. There is a relationship between low grip strength and reduced BMD on both spine and femoral neck, as well as an increased risk of incident vertebral fracture in women [9]. Axial bone loss associates robustly with the improved age-grouped grip strength quartile among postmenopausal women [10].

Positive results for grip strength and site specific bone mass are found in adolescents [11–14]. The systemic relationships between grip strength and other muscle strength such as leg power, and distant bone are also demonstrated in adolescents [15–17].

However, only a few studies have been conducted on the evaluation of the possible role of confounding variables, which may be substantially related to both BMD and muscle strength. Physical activities, calcium intake, anthropometric parameters and pubertal stages have long been recognized as main determinants of bone accretion in adolescents [18–22]. Confounding variables such as anthropometric variables may contribute to the disputing results concerning the relationship between grip strength and BMD [23]. The problem of confounding factors is more serious in adolescents because of the large variability in body size due to varying degrees of pubertal development.

Moreover, a majority of the studies were conducted in white adolescents to establish the association between muscle strength and bone mass. There is uncertainty when extrapolating from these studies to Asian adolescents because of substantial differences in peak bone mass and body size in Caucasians when compared with their Asian peers [24].

Therefore, the focus of this study was to evaluate the magnitude of the correlation between grip strength and bone mass at distant bone in Hong Kong adolescents. Potential confounders such as weight, height, pubertal development, weight-bearing activities and calcium intake were controlled in the statistical analysis.

Subjects

This cross-sectional study was part of a longitudinal study (The Hong Kong Adolescent Bone Health Cohort Study) conducted by Jockey Club Centre for Osteoporosis Care and Control, the Chinese University of Hong Kong. Baseline measurements taken from November 2003 to October 2004 were reported in this study. Hong Kong Chinese adolescents, 11- to 12-year-old boys and 10- to 11-year-old girls were included. The choice of age group was planned to capture early puberty adolescent. A total of 169 boys and 173 girls were recruited from nine primary schools geographically spread throughout Hong Kong. All subjects were normally growing adolescents who were free from chronic disease and had received no prior therapy with a known effect on bone metabolism. The study was approved by the Ethics Committee of the Chinese University of Hong Kong, with informed parental consent and subject assent of all participants.

Methods

Anthropometric and bone mass measurement

Weight (kg) and height (cm) were measured before bone mass measurement. Using the Physician Balance Beam Scale (Healthometer, Illinois, USA), we measured body weight to the nearest 0.1 kg, with subjects wearing a light gown. Height was measured to the nearest 0.1 cm with subjects barefoot by Holtain Harpenden stadiometer (Holtain Ltd, Crosswell, UK). Weight and height were used instead of body mass index (BMI) as they have stronger effect in the calculations.

Bone mineral density (BMD, g/cm2) and bone mineral content (BMC, g) at the total hip of left side (femoral neck, intertrochanteric area and Ward’s triangle), spine (L1–L4) and whole body were measured by dual X-ray densitometry (DEXA) (QDR Model 4500W, Hologic Inc, Waltham, MA). Phantom calibration was done everyday. Using the low density mode, we had considerable experience with the assessment of BMD in adolescents [25]. In our laboratory, the CV for BMD measurements were within 1.5% at all sites with subject repositioning.

Bone mineral content and bone mineral density are areal measurements, which are highly influenced by somatic growth and bone size of the scanned sites. Longitudinal studies for growing adolescents use BMC as outcome measures because subjects’ body size changes during the rapid growth period [26, 27]. For cross-sectional studies, BMD can be used since it is not required to consider the individual change in body size. However, short adolescents with or without normal bones will have a lower BMC than other healthy age-matched peers. It is also difficult to interpret areal BMD for adolescents with short stature as the similar situation of BMC. Areal BMD calculated as BMC/(bone area) is measured with anterior-posterior osteodensitometry represents a mixture of areal density and skeletal size. Because BMD does not correct for the thickness of bone, there may be an underestimation of areal bone mineral in smaller individuals and an overestimation in larger subjects [28]. So the problem of size effect in bone densitometry on adolescents should be considered. An appropriate way to correct results for body size is to use multiple regression analysis simultaneously to adjust BMC for weight, height, and other relevant factors such as age, pubertal status, and calcium intake [29]. By including body parameters, for examples, weight and height in the prediction equations, the prediction of BMC and BMD can be significantly improved [30]. As a result, our study design can reduce the effect of bone size on bone measurement.

Handgrip strength measurement

Handgrip strength was measured by Jamar Hard dynameter (Sammons Preston, Canada), and the results were expressed in kilograms. The measurements were made with the subjects sitting and their arms placed straight by their side. Subjects should grip dynamometer as hard as possible for 3 seconds. At the same time, they should not press the instrument against their bodies or bend at the elbow [31]. The subjects were given verbal support to generate maximal effort. The peak force recorded by the dynamometer was used to represent each subject’s maximum handgrip strength. Both hands were tested three times and the highest value was used. In order to identify the dominant hand of the subjects, they were asked to indicate which hand they use for writing.

Assessment of pubertal status

The Tanner grading system was used to assess sexual development by the researcher of the same sex [32]. For accurate Tanner ratings, it is suggested that they are carried out by health professionals [33]. With the application of the method of Tanner, pubertal development was evaluated by researchers’ assessment of breast and pubic hair stage in girls and genitalia and pubic hair stage in boys. Researchers were given pictures and written descriptions best reflecting the subjects’ appearance. When there were discrepancies among criteria, greater emphasis would be placed on the degree of breast development in girls and, testicular and penile size in boys in determining the Tanner stage.

Calcium intake assessment

Each subject was interviewed separately by an experienced dietitian. Some of their parents assisted in answering since subject’s meals were prepared by them. A food frequency method was used for dietary assessment of average daily calcium intake in a year. The validity of the method was examined [34]. We had extensive experience with the application of this method in Chinese adolescents [35]. The food processor nutrition analysis and fitness software V8.0 (Esna Research, Salem, USA) was used to analyze the data. We adapted this to local use by adding the composition of local food to the list, using the relevant food composition tables [36].

Weight-bearing exercise assessment

Participants were required to make a list of different organized sport teams which they had joined within past year. All sport teams were classified to the four categories: (1) the physical activities with GRF greater than four times body weight (score 3); (2) those with GRF values between 2 and 4 (score 2); (3) those with GRF values between 1 and 2 (score 1); and (4) those with GRF values of 1 (score 0); which were based on ground reaction forces (GRF) to determine the mechanical component of sport team [37]. Those subjects who reported no participation on any sport teams were assigned a zero. Mechanical components of physical activities (MECHPA) were then obtained by putting all scores altogether. It was unrelated to the duration, frequency and metabolic intensity as supported by Van Langendonck [38]. Van Langendonck suggested that the type of sport participation is more important than the duration of the participation of sports. Calculation of weight-bearing activities by MECHPA was validated, showing an association between self-reported weight-bearing physical activities and lumbar BMD in adolescents [39].

Statistical methods

Statistical analyses were performed separately for the two sexes to allow for sex differences and different measures of the pubertal stages. Pearson correlation coefficients were used to evaluate univariate relationships between bone mass parameters at different sites and other continuous variables. Spearman correlation coefficients were calculated for pubertal stages. To investigate the independent predictors of bone mass, BMC and BMD of each site were used as a dependent variable among the explanatory variables using multiple regression models. Ordinal variables were used for pubertal stage, each measuring the change from one pubertal stage to the next. Only one ordinal variable was used as linear trend of pubertal stages against bone mass was found. The predictors of BMC and BMD were quantified by linear stepwise regression analysis. Two-tailed p-values at level of 0.05 or less were considered to be of statistical significance. The SAS 9.1 was used.

Results

Table 1 describes the main features of the considered variables in boys and girls. For each variable, the mean, standard deviation and range are presented. All continuous variables of boys were significantly greater than those of girls, at the same time, more boys had advance pubertal stages (p < 0.05).

Table 1 Mean, standard deviation and range of all studied variables in boys and girls
Full size table

The positive correlations between BMC of different sites and grip strength are shown in Fig. 1. The correlations of BMC and BMD at hip, spine and whole body with various measures in boys and girls are shown in Table 2. The results were similar for both sexes and the correlation coefficients for BMD were less than BMC at all sites. There were significant correlations between grip strength and bone mass at different sites (boys: BMC: hip = 0.74, spine = 0.72 and whole body = 0.73; BMD: hip = 0.48 and spine = 0.60; girls: BMC: hip = 0.72, spine = 0.71 and whole body = 0.72; BMD: hip = 0.61 and spine = 0.63; p < 0.0001). However, grip strength was less correlated to the BMD of whole body (boys : 0.38; girls: 0.44; p < 0.0001). The coefficients between grip strength and BMC were markedly higher than those of BMD. The results of weight and height were similar to that of grip strength in both sexes. Pubertal stages were significantly related to bone mass but there were small correlation with hip and whole body BMD in boys (boys : BMC = 0.45–0.49; p < 0.0001; BMD at hip, spine and whole body = 0.27, 0.43 and 0.21; p < 0.01; girls: BMC = 0.60–0.67, BMD = 0.39–0.56; p < 0.0001). Weight-bearing activities of boys correlated significantly in moderate magnitude, the results were consistent with BMC and BMD (r = 0.27–0.35; p = 0.0001–0.0026). For girls, there were lower correlations and an insignificant association with whole body BMD (r = 0.18–0.25; p = 0.0008–0.0211; whole body = 0.08; p = 0.3213). The correlations between weight bearing activities and grip strength for boys and girls were 0.3205 and 0.2476, respectively (p < 0.01). Small correlations for calcium intake with bone mass at all sites were found in both sexes (r = 0.07–0.19; p = 0.0271–0.3781).

Fig. 1
figure 1

Relationship between grip strength and BMC results in different sites. Plus indicate results for boys, circles are results for girls

Full size image
Table 2 Pearson’s correlation coefficients of BMC and BMD at hip, spine and whole body with various measures
Full size table

BMC and BMD of each site were used as dependent variables with grip strength among the predictors to figure out the independent predictors of bone mass. The estimations were along with weight, height, weight-bearing activities, calcium intake and pubertal stages via the approach of a multiple regression model. The results of multiple regression analysis are shown in Tables 3 and 4. Six factors explained about 70% and 50% of the variations in BMC and BMD at all sites, respectively, except for whole body BMD. Grip strength was an independent predictor of bone mass except hip BMD in boys and whole body BMD in girls. For boys, weight and weight-bearing activities were independent predictors of bone mass except whole body BMD. Height was significant in BMC but not in BMD. The results of girls were similar to those of boys, except the independent role of weight-bearing activities was significant at spine only.

Table 3 Multivariate analysis of dominant handgrip strength with the determinants of BMC and BMD at hip, spine and whole body in boys
Full size table
Table 4 Multivariate analysis of dominant handgrip strength with the determinants of BMC and BMD at hip, spine and whole body in girls
Full size table

Stepwise regression model (Table 5) showed that grip strength was a robust predictor of bone mass in both sexes with an exception for hip BMD in boys. Table 6 shows the prediction model of bone mass by grip strength and weight. For both sexes, grip strength and weight could be used to explain approximately 60% of the variations in BMC of different sites and 40% in BMD of hip and spine.

Table 5 Stepwise regression model of dominant grip strength with the determinants of BMC and BMD at hip, spine and whole body
Full size table
Table 6 Multivariate analysis of dominant grip strength and weight of BMC and BMD at hip, spine and whole body
Full size table

Discussion

Strong positive correlations were shown between grip strength and bone mass at various sites of adolescents in the present study. Multiple regression analysis showed that grip strength was a strong independent predictor of BMC and BMD at hip, spine and whole body.

The results supported the mechanostat theory that muscle strength is related to bone mass during youth. It postulates that developmental changes in bone strength are secondary to the increasing loads imposed by larger muscle forces [40, 41]. It predicts that the increasing muscle mass during development creates the stimulus for the increase in bone mass. Muscle force during development varies directly with bone strength, then the increase in muscle development must come before and should determine the increase in bone mass.

Unfortunately, the analysis of both site-specific and systemic relationship in the current study was not allowed as the radius bone density measurement was absent. However, the aforementioned association between grip strength and bone mass at radius might suggest the relationship between grip strength and bone mass was both site specific and systemic.

It was interesting to find that grip strength was related to bone mass at all sites measured. During normal activity, large joint reaction forces are produced by skeletal muscle contraction forces. Therefore, the above-mentioned relationship between muscle strength and bone mineral density of nearby skeletal structures should not be surprising. This study had reported the significant relationship between grip strength and BMD of distant skeletal sites. This indicated a more perplexing relationship between muscle strength and bone mass than that of direct force by muscles on bone. Several explanations of systemic relationship between grip strength and bone mass at distant skeletal sites were formulated. Firstly, a similar finding had been found in young women in whom dominant grip strength had acted as an independent predictor of spine BMD [7]. This relationship, which was inferred by arm activity, had been linked to the simultaneous contraction of trunk-stabilizing muscles that directly exerted forces on the spine. Also, Bevier explained that the results of his study that grip strength significantly predicted spine bone density in women because grip strength and back strength themselves were significantly correlated [6].

However, the role of grip strength as an independent predictor of bone mass still existed in the non-dominant hand in the present study, even though there were significant differences in grip strength between dominant and non-dominant hands (p < 0.0001 in paired t-test). The result showed that the relationship between muscle strength and BMD was independent of regular activities by the dominant hand. It signified a general association rather than a local cause-and-effect relationship. In the studies of middle-aged men [42], the systemic association between muscle strength and bone mass was due to the more favorable biochemical profile of the subjects with good physical fitness, including low levels of parathyroid hormone [43]. A general relationship between muscle strength and BMD is possibly grounded in a simultaneous and equivalent increase in these two parameters in adolescents with a relatively low level of physical activity [17].

Lastly, because of the indication of twin studies, genetic factors may be said to decide the muscle mass and bone [44]. The growth of muscle and bone were hypothesized to be determined independently by genetic mechanisms. It may be acknowledged that the present data did not establish a direct cause-and-effect relationship between muscle force and bone mass, i.e., whether the relationship between muscle strength and bone density were due to common genes or a proportional gain in bone density caused by improved strength. In this view, prospective studies should be implemented in order to assess the changes of bone mass during strength training.

The correlation coefficients between grip strength and BMC were markedly higher than those of BMD. In growing girls, a greater muscle area was associated with a greater bone cross-sectional area [13]. The correlation coefficients for BMD were less than those of BMC as bone area was adjusted as a result of the close relationship between muscle strength and muscle area. The higher values of correlation in BMC were similar to those of another studies in girls [11,16].

For boys, our results showed that weight-bearing activities were independent predictors of bone mass except whole body BMD. But the independent role was only significant at spine for girls in multivariate analysis. It might be due to lower participation in weight-bearing exercise in girls. The systemic association between weight-bearing exercise and bone mass suggests that there is a mediated effect throughout the skeleton. Bone metabolism assessed by serum markers demonstrated favorable systemic effects of physical exercise [43]. Therefore, because of the physical stress, localized enhancement of bone formation can probably lead to favorable effects on bone metabolism in the rest of the skeleton.

The small values of univariate correlation in calcium intake were similar to those of other observational study in adolescents [22]. Calcium supplement would have more effect for intervention studies when considering BMD change [19]. Non-significant result in multivariate analysis of this study might due to the effect of calcium intake was not strong enough after adjusting other covariates such as weight and grip strength. Moreover, weak association might due to narrow distribution and low consumption of calcium intake as the true influence of calcium in bone could not be shown from those adolescents. Mean calcium intake was 619.51 ± 242.78 mg/day and 569.07 ± 274.16 mg/day among boys and girls, respectively, in this study. Subjects had low calcium intake compared to recommendation of 800–1000 mg per day [45]. That may due to the non-milk drinking habit in Chinese. The beneficial effect may be established if sufficient calcium intake is available. On the other hand, recall bias might exist because adolescents were required to report their dietary intake in the past year.

Pubertal development is a key determinant of bone accretion in adolescents. However, the correlations among pubertal stages and bone parameters were moderate and only significant in spine, but not in hip and whole body in multiple regression analysis. The reason might be the narrow range of pubertal stages in our subjects, as most of them were early puberty.

Our results were consistent with the majority of those from previous studies showing body weight was a strong independent predictor of bone mass in adolescents. Analysis of association between bone mass and any variable without adjusting weight will be hampered by it. From our results, grip strength was still a significant predictor of bone mass after adjusting weight, meaning that grip strength was a robust and strong predictor for bone mass. Height was significant in BMC but not in BMD because BMD was already adjusted by body stature.

Since grip strength and weight were the most important predictors for bone mass, a combination of those parameters may provide high predictive power for bone mass. For both sexes, grip strength and weight could be used to explain about 60% of the variance in BMC of different sites and 40% in hip and spine BMD. The result showed that it might be sufficiently strong to permit measurement of grip strength plus weight to be used for prediction of the bone mass in early puberty period. It was preliminary analysis hence further diagnostic studies by receiver operating characteristic (ROC) curve were required in future.

Based on the fact that adolescence is the critical life period for bone mineral accrual, the method to identify pediatric patients with skeletal compromise should be reliable and essential for the development of prevention strategies. The ideal measurement of the pediatric bone mass should be safe and readily available. It is also easy to perform with adolescents of all ages. DEXA is patient-friendly, highly precise, and involves only minimal radiation exposure, but it is bulky and expensive. Measurement of grip strength is simple, neither expensive nor invasive. Although the actual uncertainty is about its basis, significantly correlated grip strength among different sites of bone mass may be considered as an additional factor to the factor “weight” to be a predictive method for detecting bone structure in early puberty period. Even it is not an adequate substitute for bone densitometry, the risk groups at which to direct bone density measurement can be identified. Intervention such as weight-bearing exercise and calcium supplement can be introduced earlier in order to let adolescents increase bone mass optimally before bone growth spurt.

This study is cross-sectional with all its limitations which reflects associations but not reveals causes and effects. Recall bias might exist in calcium assessment because adolescents were required to report their dietary intake in the past year.

Conclusion

We conclude that grip strength is an independent predictor of bone mineral density in both sexes. Relationship of muscle strength to bone mass is not only site specific but also systemic. Prediction model by grip strength and weight may be used for prediction of the bone mass in early puberty period.