Weight-bearing exercise and calcium intake are known to contribute to bone density. However, the relative significance of physical activity and calcium intake in the development of bone characteristics in functionally different weight-bearing and nonweight-bearing bone sites at different ages is poorly known. A total of 422 women in three age groups (25–30, 40–45, and 60–65 years) were screened from 1017 women and divided into four groups by their level of physical activity (high [PA+] and low [PA−]) and calcium intake (high [Ca+] and low [Ca−]). Total body bone mineral content (TBBMC), areal bone mineral density (BMD) of the femoral neck and distal radius, and selected dimensions and estimated strength variables (bone width, cortical wall thickness, cross-sectional moment of inertia, and section modulus of the femoral and radial shafts) were measured with dual-energy X-ray absorptiometry. Both high physical activity and high calcium intake were associated with a higher TBBMC when compared with low activity and calcium intake (1.8% and 4.6%, respectively). The BMD of the weight-bearing femoral neck was 5% higher in the PA+ groups than in the PA− groups, whereas calcium intake showed no such significant association. Neither physical activity nor calcium intake was associated with the BMD of the nonweight-bearing radius. However, both high physical activity and high calcium intake were related to larger and mechanically more competent bones in the femoral and radial shafts, the association for physical activity being stronger with increasing age. No significant interaction between physical activity and calcium intake was found with respect to any of the bone variables. These data from a cross-sectional study suggest that a moderate level of physical activity or a sufficient level of calcium intake, if maintained from childhood, can result in considerable long-term improvement in the mechanical competence of the skeleton. The clinical relevance of these findings is further emphasized by the fact that the observed patterns of physical activity and calcium intake pertain to customary lifestyle and are thus feasible targets for the primary prevention of osteoporosis.
Physical activity probably contributes to increased bone density, especially in childhood and youth, and also probably helps to attenuate age-related bone loss.(1) Several studies have reported a high bone density in physically active subjects as compared with that of sedentary subjects in both cross-sectional(2-5) and prospective studies on premenopausal(6-8) and postmenopausal(9) women. The highest densities have been found in bones subjected to high-magnitude or high-impact loading such as in weight lifting and racket games.(1,3) However, the role of customary physical activity has remained inconclusive. A positive association with bone mineral density (BMD) has been reported for postmenopausal women(10) but not for premenopausal women with customary physical activity.(3,11)
Observations regarding the relationship between calcium intake and bone mass are not consistent either.(12,13) High calcium intake has been reported to be associated with increased bone density in intervention and longitudinal studies on children(14,15) and adults,(16-20) whereas some longitudinal(21,22) and cross-sectional studies(23,24) found no association between calcium intake and bone mass. Some cross-sectional(2,25) and intervention(26-28) studies have provided evidence of beneficial effects of both calcium intake and physical activity on bone density.
Despite the central role of low bone density as a determinant of bone fragility,(29,30) patients with hip fractures have only slightly lower bone density than the age-matched subjects without fractures.(31) These circumstances comply with the fact that not only density, per se, but also the gross geometry and architecture of bone, in conjunction with intrinsic material properties, are important determinants of the mechanical competence of the bone in question.(30,32)
In line with the fact that bone is dynamically remodeled throughout life, and is adapted to its mechanical loading environment,(33) bone geometry is altered while bone mineral loss follows an apparent age-related pattern.(34) Consequently, geometric changes in bone might compensate for the adverse effect of bone loss without substantial reduction of bone strength. This issue has remained controversial, however, and is probably site- and gender-specific.(35-38) It has been suggested that assessing the gross geometry of bone in conjunction with standard measurements of bone density could provide a better description of the mechanical competence of bone(35,39) and thus improve the accuracy of predicting fracture risk in general.(40,41)
We undertook the present study to determine the contributions of physical activity and calcium intake to bone mass and size, and thus to bone strength, among healthy women of different ages. We paid special attention to potential differences between the weight-bearing and nonweight-bearing bones and also between the long bone epiphyseal and shaft sites, which exhibit distinct differences in cortical to trabecular bone ratios.
MATERIALS AND METHODS
Study design and subjects
Using a newspaper advertisement, we recruited healthy women in the age groups of 25–30, 40–45, and 60–65 years. The rationale behind this age division was the possibility of an age-specific association between physical activity and calcium intake among women whose bone mass is at or close to maximum, in apparent steady state, and those whose postmenopausal bone loss has leveled off. We also divided the subjects into four groups according to current physical activity and calcium intake to obtain two physically active groups (PA+) with different calcium intake, one being abundant (Ca+) and the other being low (Ca−), and two physically inactive groups (PA−) with the same contrast in calcium intake. Our preliminary calculations suggested that to be able to detect a clinically important (5%, estimated by us) and significant (p < 0.05) intergroup difference in BMD with a statistical power of 80%, ∼35 subjects would be needed for each group.
A total of 1017 women were willing to participate and were subsequently screened with the use of a diary to assess calcium intake(42) and a detailed pretested questionnaire to assess health status and habitual physical activity. The criteria for selection were a current dietary calcium intake of over 1200 mg/day (Ca+, higher intakes are not warranted as shown by supplement studies(43,44)) or under 800 mg/day (Ca−, RDA recommendation for adult women), and a current physical activity regimen of vigorous activity more than twice a week (PA+, at least 20 minutes per session causing enhanced breathing, e.g., jogging, aerobics or participation in moderate-to-high intensity racket games) or moderate or light activity no more than once a week (PA−, e.g., walking causing only a light elevation of heart rate). Women who had intermediate physical activity or calcium intake levels or who used calcium supplements were excluded. Estrogen replacement therapy was also an exclusion criterion for the oldest age group if it had been used during the 7 years preceding the study or if used continually for a period of at least 3 years after menopause. Consequently, 425 women were invited to attend the initial examination for measurements; three of them had to be excluded, however, because of incomplete data. The remaining 422 subjects represented the three age groups as follows: 132 women in the youngest group, 157 women in the middle-aged group, and 133 women in the oldest group. The number of subjects in each group is shown in Table 1. All the subjects were nonsmokers and nonobese (body mass index [BMI] < 28). None of the subjects had diseases or injuries or used drugs that could have affected bone or calcium metabolism, except the potential brief usage of estrogen during menopause in the oldest age group.
Table Table 1. Characteristics of the Study Groups
The study protocol was approved by an independent ethical committee for clinical investigations, and each subject gave their written informed consent.
Information on the participants' health, history of physical activity from childhood to present, education, occupations, work history and work load, number of children, use of hormonal contraceptives, menstrual, and menopausal status, use of alcohol, possible earlier smoking history, and use of medication, including the possible use of estrogen therapy during menopause, was collected in an interview.
The height, weight, and BMI (weight in kilograms divided by height in square meters) of each subject were measured during the visit.
In addition to the preliminary assessment of dietary calcium intake, the current total dietary intake, including calcium intake and use of dietary supplements, was assessed by a complete 4-day food record (3 weekdays and Sunday). Such a combination of two methods has been proposed as a practical and valid approach to estimating habitual nutritional status.(45) The food composition data were calculated with validated Micro-Nutrica software (Social Insurance Institution, Helsinki, Finland). Milk consumption history was recorded as the use of milk before school age, during school age, during the teenage years, and during adulthood. If milk was not currently in use, the age at which the subject stopped using it was recorded.
Each subject's current daily walking distance was measured during 3 days (2 weekdays and Sunday) with a pedometer (Fitty-3 Electronic, Uttenreucht, Germany) and heart rate was measured simultaneously with an electronic heart rate meter (Sport Tester PE, Polar Electro, Oy, Finland). In addition, the subjects kept an activity diary during these days. In it they evaluated the intensity of their concurrent activity for every half-hour on a scale from 1 (sleeping or resting) to 10 (exhausting work or exercise). The activity level was calculated as the mean of the half-hour scores during the 3 days and was used as the indicator of current physical activity. Information on the subjects' lifetime physical activity was obtained with a standardized questionnaire. Physical activity from childhood on was classified into four categories according to type and frequency as high, moderate, low, and no activity, high being vigorous activity at least twice a week (at least 20 minutes per session causing enhanced breathing and elevation of heart rate), moderate being vigorous activity once a week, and moderate activity a few times a week, low being moderate activity once a week or light activity several times a week (e.g., walking causing no enhanced breathing and only light elevation of heart rate) but no vigorous activity, and no activity being no regular physical activity. The years in different classes was obtained from the age of 16 years on. In the analysis, the high and moderate classes and the low and no activity classes were combined. Furthermore, the subjects were asked about possible types of exercise (e.g., ballet dancing, sport exercise) and the number of years of participation in childhood before the age of 16 years.
Lifetime occupational physical activity was assessed in an interview with regard to the physical activity level during work, corresponding to multiples of the basal metabolic rate (MET).(46) Very light and light physical activity corresponded to ∼2.25 MET, moderate activity to 5.0 MET, and heavy and very heavy occupational activity to 9.0 MET. The MET factor was then multiplied by the number of years spent in each occupational activity class and divided by the total sum of years worked to obtain a work-load index.
The maximal isometric muscle strength of the trunk extensors and flexors, the leg extensors, and the dominant forearm flexors was measured by strain gauge dynamometers.(47)
Cardiorespiratory fitness (estimated maximal oxygen uptake, VO2max) was assessed by the 2 km walking test.(48) In a randomly selected subgroup of 28 subjects, the 2 km walking test was validated against a maximal bicycle ergometer test using direct ventilatory gas analysis (Sensormedics 2900Z, Sensormedics Inc., Anaheim, CA, U.S.A.). The mean difference between the walking and ergometric tests was 1.12 (4.49) ml/kg/minute, and the total error of the walking test was 4.55 ml/kg/minute.
The total bone mineral content (TBBMC, g) of the body, and the areal bone mineral density (BMD, g/cm2) of the femoral neck and distal radius of the dominant side were measured with dual-energy X-ray densitometry(39) (DXA; Norland XR-26, Norland, WI, U.S.A.). As we have done previously,(49,50) for estimating the dimensions and mechanical competence of the long bone shafts, the bone width, cortical wall thickness (CWT), cross-sectional moment of inertia (CSMI), and section modulus (Z) were estimated for the dominant femoral and radial shaft sites using simple engineering formulas developed for DXA analyses.(39) The CSMI corresponds to the rigidity (stiffness) of a long bone segment and the Z to its bending strength. All the above DXA measurements were controlled for differences in subjects' heights(39) in such a way that skeletal regions that were anatomically the same were measured irrespective of height. The average in vivo precisions of these measurements were: <2%, BMC; <1%, BMD; <1%, dimensions; and <2%, mechanical estimates.(39)
The mean and standard deviation (SD) were used as descriptive statistics. In addition, mean t-scores were calculated using the data of the young PA+Ca+ group to allow relative intergroup comparisons.
As the primary analysis of this study, the associations of physical activity and calcium intake with the bone variables were analyzed by analysis of covariance models in which physical activity and calcium intake were used as factor variables. Age was included in the model as a third factor variable to determine the age-specific associations between the variables. Body weight, body height, BMI, and work load index were considered as possible confounding variables. Of those, body weight had the highest correlation with the bone variables. After taking the body weight into account, the other confounders were not correlated with bone variables and/or factor variables, and body weight was used as the only covariate in the model for adjusting the effect due to the body size. All percentage differences in results are adjusted for body weight. To reduce the possible problem arising from multiple testing, we limited the number of outcome variables before the study and restricted them to descriptors of appropriate functional elements of the skeleton. Moreover, for multiple comparisons between the age groups, the simultaneous 95% confidence intervals were calculated by the Scheffé method.
The group characteristics are described in Table 1. The cardiorespiratory fitness (estimated VO2max) of the physically active subjects was 12% higher than that of the inactive subjects. The current daily walking distance was longer, and the daily activity score was higher in general. The muscle strength of the physically active groups was also higher than that of the inactive groups. Current activity was strongly associated with a history of physical activity. The years of vigorous activity of the PA+ and PA− groups were 10.4 and 3.4 years in the youngest age group, 20.3 and 7.0 years in the middle-aged group, and 34.6 and 13.7 years in the oldest age group, respectively. In contrast, the years of light or no activity were 1.5 and 9.1 years, 6.6 and 19.5 years, and 12.5 and 32.2 years in the PA+ and PA− groups, respectively.
The mean calcium intake of the Ca+ groups was twice as high as in the Ca− groups, 1475 mg compared with 638 mg, respectively. Almost all (94%) of the subjects with a low current calcium intake had consumed milk in childhood. The respective percentage in the Ca+ group was 98%. There were no differences between the groups with respect to the health status, education, number of children, use of hormonal contraceptives, menstrual and menopausal status, use of alcohol, smoking history, or estrogen replacement therapy during menopause, except for the history of physical activity and the work-load index, which were higher in PA+ than in PA− groups.
Bone mass and areal density
The observed absolute values of the bone variables together with the mean t-scores relative to the young PA+Ca+ group are given in Table 2. Since there was no significant interaction between physical activity and calcium intake with respect to any bone variable, the associations between physical activity and calcium intake and the bone variables have been described separately. The work-load index did not correlate with the bone variables.
Table Table 2. Bone Characteristics of the Study Groups
Total bone mineral content:
In general, the TBBMC was associated with physical activity and calcium intake (Fig. 1). The mean value was 1.8% (95% confidence interval, 0.1–3.7) higher in the PA+ than in the PA− groups. There was no significant difference between the PA+ age groups. The association between calcium intake and TBBMC seemed to increase with age (PCaxAge = 0.096), with no difference (−0.5%) being found between the Ca+ and Ca− groups in the youngest age group, and a difference of 4.6% (0.5–8.9) being recorded for the oldest age group (Fig. 1).
Areal bone mineral density:
A higher level of physical activity was strongly positively associated with the BMD of the femoral neck (PPA < 0.001) in all the age groups, the association showing no difference between the age groups. The mean value of the femoral neck was 4.7% (2.3–7.1) higher in the PA+ than in PA− groups (Fig. 1). Calcium intake was not related to the femoral neck value. Neither physical activity nor calcium intake was significantly related to the BMD of the radius in any age group (Fig. 1).
The association between femoral bone width and physical activity was age related (PPAxAge = 0.003). In the oldest age group, the mean difference between the PA+ and PA− groups was 5.2% (1.7–8.8). In the other age groups, the differences were not significant (Fig. 2). In contrast, in the Ca+ groups, the femoral shafts were 1.6% (0.1–3.2) wider than in the Ca− groups, the association showing no difference between the age groups.
For the radial shaft, the association between physical activity and bone width was age related (PPAxAge < 0.001), the mean difference being −1.3% in the youngest age group, 1.9% in the middle-aged group, and 5.7% in the oldest group. The radial shafts of the Ca+ groups were 1.6% (0.5–3.3) wider than in the Ca− groups, the association showing no difference between the age groups (Fig. 2).
Cortical wall thickness:
The CWT was thinnest in the older age groups in both the femoral and radial shafts (Table 2), regardless of the level of physical activity or calcium intake. The associations with physical activity or calcium intake were not significant for the femoral shaft (Fig. 2). The mean CWT of the radial shaft of the PA+ groups was 2.2% thinner than that of the PA− groups (−4.3–0.0), but the difference was not age related. In contrast, the association with calcium intake was age related (PCaxAge = 0.041). The radial CWT of Ca+ group was 5.4% wider than in the oldest Ca− group (0.3–10.7) (Fig. 2).
Estimated bone strength variables
Cross-sectional moment of inertia:
The values of the CSMI increased with age in the PA+ groups, but decreased in the PA− groups (Table 2). For the femoral shaft, the association with physical activity was age related (PPAxAge = 0.002), the difference being largest (18.1%, 7.9–29.3) in the oldest age group. The mean difference between the Ca+ and Ca− groups was 5.4% (1.1–9.9), and it showed no age-related association (Fig. 2).
For the radial shaft, the association between the CSMI and physical activity was age related (PPAxAge = 0.002), the youngest group exhibiting a −4.5% mean difference, the middle-aged group a 6.9% mean difference, and the oldest age group a 17% mean difference. The association with calcium intake was not age related, even though the mean difference was greatest in the oldest age group (12.7%). The mean radial CSMI of the Ca+ groups was 7.4% (2.7–12.3) higher than the respective value of the Ca− groups (Fig. 2).
The association of physical activity with Z of the femoral shaft was age related (PPAxAge = 0.003), the mean difference between the PA+ and PA− groups being significant in the oldest age group only (12.3%, 5.4–19.6). The association between calcium intake and the femoral Z was similar in all the age groups, but the Ca+ groups had a 3.8% higher value than the Ca− groups (0.9–6.7) (Fig. 2).
For the radial shaft, the association between physical activity and Z was clearly age related (PPAxAge = 0.005), the youngest group differing from the others. The difference between the PA+ and PA− groups was −3.2% in the youngest age group, 4.1% in the middle-aged group, and 10.7% in the oldest group. Calcium intake was positively associated with the Z of the radial shaft in all the age groups, the mean difference being 5.4% (2.1–8.9). The association was not different between the age groups (PCaxAge = 0.20), even though the 10% difference in the oldest age group was larger than that of other age groups (Fig. 2).
Our results provide indirect but clear evidence of the benefits of increased levels of physical activity and calcium intake on the total BMC of the body and regional mineral characteristics of the skeleton. Both of these life style factors can modify the well-known age-related bone mineral loss and periosteal expansion of long bone shafts that occur simultaneously with the thinning of the cortices. As a result, skeletal integrity is improved in women who are physically active and have a high calcium intake. It is noteworthy that the benefits induced by physical activity and calcium intake as suggested by our results had characteristic site-specific manifestations of their own and they followed different time courses. However, this study revealed no significant interactions between physical activity and calcium intake.
The interpretation of our results stipulates that most of the subjects who reported being currently physically active had been active throughout their lives. The results of an earlier study(51) suggest that long-term recall of physical activity is reliable. Furthermore, the subjects who reported having a high current calcium intake had had a high calcium supply throughout their lives as well. Respective strong correlations between the current situation and the subjects' history and the observed consistent differences in the physical performance of the subjects together corroborate our assumptions and support the interpretation that our findings show the lifelong influence of physical activity and calcium intake on bone.
The age-independent consistency of differences in BMD in relation to physical activity implies that the net benefit of physical activity for the skeleton can be obtained during development and growth and maintained thereafter. Extrapolation from data of recent exercise intervention trials(7-9) indicates that the required period with which to obtain a significant gain is probably at least a year. Furthermore, this difference, which amounts to about 5%, agrees with those observed for young endurance athletes,(3-5) who had mainly started systematic training during their peripubertal years. Whether it is possible to obtain increases in BMD of this magnitude at any age during adulthood within a given period remains unclear. Conclusive evidence of this possibility from sufficiently long controlled exercise trials is lacking. What is known is that the effect of physical activity is the most pronounced during the growth spurt at puberty(52) and that about 5% differences in BMD are observed between competitively and normally active girls already at that age.(53)
Our work-load index and measure of light leisure-time physical activity were not associated with the bone variables. This observation agrees with the results of our earlier studies, in which neither occupational weight-bearing loading in adulthood among occupationally active middle-aged healthy women nor normal-level activity among young women seemed to benefit the skeleton substantially.(3,11) This finding was probably attributable to a loading-induced stimulus that is insufficient to increase bone formation. In any case, had the subjects' skeleton benefited from physical activity practiced for any reason, the maintenance of the acquired benefit should require that the relative intensity and type of loading be continually maintained. It is known that the skeletal benefits of nonathletic physical activity are lost if loading is discontinued.(54,55) However, recent cross-sectional observations on the bones of former competitive athletes(4) indicate that some of the benefits of extreme bone loading begun at a young age and continued for a long time can remain, even when physical activity is continued at a considerably lower intensity than during training aimed at competitive sport.
The observation that the average calcium intake of the Finnish population meets current recommendations,(56) even among those who avoid consuming fresh milk, may well have modified the observed calcium-related associations. Actually, it was difficult to find subjects with a low calcium intake, and this difficulty led to the lower than planned sample sizes of some of the Ca− groups. As regards the crucial peripubertal years in terms of skeletal development, virtually all of our subjects (397 out of 422) had consumed milk at that time. Most of the 25 subjects who had not consumed milk during school age were in the youngest group, whereas the other subjects with a low calcium intake had substantially decreased their use of milk between the ages of 25 and 40 years. This reduction in calcium intake before menopause and consequent bone loss may have led to the observed differences in the bone variables of the older subjects with high and low calcium intake. In a population in which the average lifelong calcium intake is low, the role of calcium would probably be accentuated as compared with our findings.
For the weight-bearing proximal femur, the contribution of physical activity was consistent across the age groups, whereas calcium intake showed no significant association. Neither did the nonweight-bearing distal radius show a significant association with calcium intake or physical activity in any group. In contrast, the trend for TBBMC was similar to the femoral neck findings and significant both in terms of physical activity and calcium intake. These observations can, at least partly, be explained by the fact that the cortical component of the skeleton dominates and therefore accounts mainly for the total body measurement. The cortical long bone shafts exhibited characteristic associations with physical activity and calcium intake, which association followed age-related patterns of their own. In addition to its apparent general benefit on the skeleton, physical activity was associated with a clear age-related widening of the shafts of the weight-bearing femur and nonweight-bearing radius as compared with the corresponding findings of the nonactive groups.
In line with the changes in gross geometry, the indices of mechanical competence (CSMI and Z) improved with age for both long bone shafts. Like physical activity, high calcium intake was also associated with larger and mechanically more competent long bone shafts in general, but in contrast showed no distinct age relation. The weight-bearing femoral shaft especially seemed to benefit from high calcium intake in terms of bone strength already at a young age. However, the overall calcium-related strengthening of the shafts was not as marked as the apparent long-term influence of physical activity.
Our design with two contrasted levels of physical activity and calcium intake in three age groups was planned to evaluate the association of these two factors with bone variables. However, we found no interaction between physical activity and calcium intake. Our design, even though cross-sectional and providing only indirect evidence, was likely to reveal potential causal relationships between skeletal health and physical activity and calcium intake. Hence, in light of the consistent observations, both regarding single long bones (ends vs. shafts) and functionally different bones (weight-bearing femur vs. nonweight-bearing radius), our observations are plausible. Furhermore, given the homogeneity of the racial, educational, and social characteristics of the Finnish population, our relatively large sample of women representing three different age groups can be considered appropriate for evaluating the associations in a healthy Caucasian female population in general, despite the fact that the subjects were volunteers. In this respect, it should be noted that the average intensity, magnitude, and frequency of physical activity in its various forms was moderate, even among our active subjects.
Our results suggest that a relatively moderate level of physical activity or calcium intake, if maintained from childhood, can result in considerable long-term benefits with respect to the mechanical competence of the skeleton at the population level. The clinical relevance of these findings is further emphasized by the fact that the observed patterns of physical activity and calcium intake of the subjects in our study pertain to customary life style and are thus valid targets for activities aimed at the primary prevention of osteoporosis.
The authors thank the Emil Aaltonen Foundation and the Yrjö Jahnsson Foundation for their financial support.