Osteoporosis in Elderly Men and Women: Effects of Dietary Calcium, Physical Activity, and Body Mass Index


  • T. V. Nguyen,

    Corresponding author
    1. Bone and Mineral Research Program, Garvan Institute of Medical Research, St Vincent's Hospital, Sydney, New South Wales, Australia
    2. Wright State University School of Medicine, Dayton, Ohio 45325, U.S.A.
    • Division of Human Biology Wright State University School of Medicine 1005 Xenia Avenue Yellow Springs, OH 45387, U.S.A.
    Search for more papers by this author
  • J. R. Center,

    1. Bone and Mineral Research Program, Garvan Institute of Medical Research, St Vincent's Hospital, Sydney, New South Wales, Australia
    Search for more papers by this author
  • J. A. Eisman

    1. Bone and Mineral Research Program, Garvan Institute of Medical Research, St Vincent's Hospital, Sydney, New South Wales, Australia
    Search for more papers by this author


Dietary calcium intake and physical activity are considered practical measures for prevention of osteoporosis. However, their associations with bone mineral density (BMD) in the elderly are not clear. The present study examined the association between osteoporosis and these two factors in relation to body mass index (BMI) in a cross-sectional, epidemiological study involving 1075 women and 690 men, aged 69 ± 6.7 years (mean ± SD). Dietary calcium intake (median of 580 mg/day) was inversely related to age (p = 0.01), positively related to physical activity index (PAI) (p = 0.01), femoral neck BMD (p = 0.01) in women, and higher lumbar spine (p = 0.003) and femoral neck BMD (p = 0.03) in men. Quadriceps strength was negatively associated with age (p < 0.0001) and positively related to BMI (p < 0.0001) and BMD (p < 0.0001) in both men and women. The PAI was associated with quadriceps strength (p < 0.0001) and femoral neck and lumbar spine BMD in women (p < 0.001) and with femoral neck BMD in men (p = 0.04); however, these associations were not significant after adjusting for age, BMI, quadriceps strength, and dietary calcium. Women in the top tertile of quadriceps strength (≥23 kg) and dietary calcium intake (≥710 mg/day) had 15% higher BMD than those in the lowest tertiles (≤15 kg and ≤465 mg/day); the difference was comparable in men (11%). Among subjects with the lowest tertiles of BMI (≤23 kg/m2 for women and ≤24 kg/m2 for men), quadriceps strength (≤15 kg for women and ≤28 kg for men), and dietary calcium intake (≤465 mg/day), 64% and 40% of women and men, respectively, were classified as having osteoporosis (based on a 2.5-SD reduction from the young-normal mean). The prevalence was only 12% in women and 1.5% in men among those in the highest tertiles of the three factors. Adequate dietary calcium intake and maintaining a physically active lifestyle in late decades of life could potentially translate into a reduction in the risk of osteoporosis and hence improve the quality and perhaps quantity of life in the elderly population. (J Bone Miner Res 2000;15:322–331)


In recent years, with the realization that bone mineral density (BMD) can be used as the primary indicator of risk of osteoporotic fractures,(1–3) considerable interest has focused on devising strategies for the prevention of osteoporosis in the elderly, a condition that is becoming increasingly prevalent because of the rapid aging of contemporary populations. BMD is a complex trait, influenced by both genetic and environmental factors and possibly interactions between them. Among the array of environmental factors, nutrition, physical activity, and gonadal hormonal factors are believed to be involved in the regulation of BMD. It seems logical that a practical strategy for prevention of osteoporosis is to improve the physical activity level as well as dietary habits in the general population. However, evidence for the association between these factors and BMD is inconsistent and sometimes confusing.

Indeed, although about 99% of calcium in the body is contained in the skeleton,(4) it has not been easy to show any definite association between calcium intake and bone mass in man. For example, results of studies examining the relationship between calcium intake and BMD have ranged from positive association(5–7) to weak relationships(8) and nonsignificant relationships.(9–13) Previous studies have not necessarily taken into account confounding factors such as weight and age. Also, because of the measurement techniques or the populations themselves, dietary calcium intake has been highly variable between studies. More importantly, many of the studies were based on relatively small sample sizes and may have not had adequate power to define modest associations between dietary calcium intake and BMD. However, there is a general consensus that dietary calcium moderately reduces the rates of cortical bone loss.(14–17)

Available epidemiological evidence, when coupled with relevant experimental and clinical research, suggests that physical activity can favorably influence the development and maintenance of bone mass and delay the progression of osteoporosis. Even a short period of immobilization or weightlessness results in rapid bone loss, with trabecular bone loss exceeding that of the cortex.(18) However, studies of the association between physical activity and bone density have not always yielded consistent findings. Although some have shown beneficial effects of exercise on bone mass in postmenopausal women,(19–24) others have not.(25–27) Most of these studies have concentrated on specific types of physical activity, notably, exercise and walking, but have not included other activities such as recreational or home-based activities, which are perhaps more important among the elderly.

Men are not exempt from osteoporotic fractures. Approximately one-third of all osteoporotic fractures in the elderly occurred in men. In fact, their lifetime risk of fractures is not that low compared with women.(28) Yet evidence is only beginning to emerge, thereby highlighting an important void in the literature concerning the distribution and determinants of osteoporosis in men. In terms of lifestyle factors, it is not clear whether dietary calcium intake and physical activity are associated with bone density in men. Nevertheless, it has been shown that higher quadriceps strength was associated with higher BMD in men.(29)

Irrespective of gender-dependent effects, study design, and measurement techniques, the major complication in any association study (among dietary calcium intake, physical activity, and bone density) is that these traits decline with advancing age and are associated with body size. The variation, covariation, and correlation among these age-related traits pose a challenge for dissecting their independent effects (if any) on bone density. Indeed, there is no reason to believe that the effect of quadriceps strength or physical activity on bone density would be independent of age or dietary calcium intake. Thus, apart from this challenge, there are a number of questions remained to be addressed, namely, (1) What is the relationship between dietary calcium intake, quadriceps strength, and BMD in relation to age and body size? (2) Is there any interaction effect between these factors on BMD? and (3) Is any relationship dependent on skeletal sites examined? The present study was designed to address these questions, by examining data obtained from an unselected sample of elderly men and women in the general community population.



The Dubbo Osteoporosis Epidemiology Study (DOES) has been carried out in Dubbo, New South Wales (Australia), a city of approximately 32,000 people, of which 1581 men and 2095 women aged 60 or above (as of June 1989) were invited to participate in the study. The population is 98.6% white. Dubbo is the base for an ongoing community epidemiological study of cardiovascular disease.(30) The age and sex distribution of population in Dubbo closely resembles the Australian population and is relatively isolated in terms of medical care, so virtually complete ascertainment of health outcomes, including fractures, is possible.(31) The study was approved by the Ethics Committee of St. Vincent's Hospital, Sydney, Australia.


After obtaining written informed consent, subjects were interviewed by a nurse coordinator who administered a structured questionnaire for anthropometric variables (height, weight) and lifestyle factors. Weight without shoes was measured at baseline and follow-up visits. Height was measured by a standiometer with precision of 0.1 cm. Body mass index (BMI) was calculated as kilograms per meter squared (i.e., weight in kilograms divided by squared height in meters.)

Assessment of dietary calcium intake was based on a 4-day weighed food frequency questionnaire.(32) Briefly, subjects were asked to provide information daily intake of milk, bread, yogurt, cheese, ice cream, eggs, fish, cereal foods, fruits and vegetables, alcohol, and others (such as chocolate and orange juice). Because milk had been found to provide the largest percentage of calcium in the diet, multiple questions relating to milk consumption were included to increase accuracy. Although the use of calcium supplements is not widespread in Australia, a question relating to their use was included in the questionnaire. The calcium content for each food item was derived from manufacturer's data when available or from Australian or British tables of food consumption. The amount of dietary calcium intake was then estimated as the sum for each of these foods from their standardized calcium content.

Quadriceps strength (maximum isometric contraction) was measured in the sitting position in the subject's dominant (stronger) leg with a horizontal spring gauge calibrated up to 50 kg of force. The test–retest reliability and reproducibility of the test have been described fully elsewhere.(33)

Physical activity was assessed at baseline as the average number of hours per day spent in each of five levels of activity and the weighting factor (WF) based on associated oxygen consumption for each activity, according to the Framingham Study.(34) The five activities were basal activity (e.g., sleeping, lying down; WF 1.0), sedentary (e.g., sitting, standing; WF, 1.1), light (e.g., casual walking; WF, 1.5), moderate (e.g., gardening, carpentry; WF, 2.4) or heavy (e.g., lifting, digging; WF, 5). The products of hours times weighting factor for all activities over a week were then summed to yield an index of daily physical activity (PAI). High PAI values correspond to lifestyles that are relatively physically active; low levels correspond to habitual inactivity.

BMD values (in g/cm2) at the femoral neck and lumbar spine were measured by dual-energy X-ray absorptiometry using a LUNAR DPX-L densitometer (LUNAR Corporation, Madison, WI, U.S.A.). The same instrument and software were used throughout the study. The radiation dose with this method is <0.1 μGy. The coefficient of reliability of BMD in our institution in normal subjects is 0.96, or 2.6%, at the femoral neck.(35)

Statistical methods

The associations among dietary calcium intake, quadriceps strength, PAI, and BMD were assessed by both univariate regression and adjusted (multiple regression) analyses. In each analysis, lumbar spine and femoral neck BMD were treated as outcome variables; dietary calcium intake, quadriceps strength, and the PAI were treated as “independent” factors; and age and BMI were treated as covariates. In the univariate analysis, each independent factor was considered in a simple regression model; in the multiple regression analysis, three sequential adjustments were considered: Age, age and BMI, and finally all variables (including the independent factors) were fitted in the same model. Estimation of the regression parameters for males and females separately was based on the least squares method. Statistical significance of these parameters was considered as p < 0.05. Verification of model assumptions such as normally independent distribution of errors and homogeneity of variances was assessed by analysis of residuals(36) by means of the SAS System.(37)

To assess whether the influences of quadriceps strength, physical activity, and dietary calcium intake on BMD were independent of each other and independent of age and BMI, second-degree interaction between the factors was also analyzed in the regression analysis. In a sense, this interaction analysis is equivalent to subgroup analysis. It is well known that subgroup analysis and categorization of continuous data often produce false-negative and false-positive results, because as more hypotheses are tested in the same data set, the set is more likely to yield some statistically significant difference even if none in fact exists. To minimize this problem, further exploratory analyses to locate underlying differences between subgroups were only performed when the p value of the interaction effect was less than 0.15. In these analyses, BMI, dietary calcium intake, quadriceps strength, and PAI were categorized into three approximately equal groups based on their tertile distribution.

To characterize the utility of these risk factors in the assessment of osteoporosis, each subject's femoral neck BMD was used to define “osteoporosis.” According to the criterion proposed by the World Health Organization (WHO), a subject is classified as having osteoporosis if his or her BMD is 2.5 SD below the young normal mean, taken as aged between 20 and 30 years old. In our sample, this is equivalent to <0.7 g/cm2 for women and <0.74 g/cm2 for men. Based on this definition, prevalence of osteoporosis was then calculated for all possible combinations of risk factors (BMI, dietary calcium intake, and quadriceps strength).

Finally, the intercorrelations among BMD, BMI, age, physical activity, quadriceps strength, and dietary calcium intake were modeled using the path analysis modeling technique. This analysis tested two specific models: (1) that BMD is related to age, BMI, quadriceps strength, and dietary calcium intake, where BMI is allowed to be correlated with dietary calcium intake and quadriceps strength is correlated with age; and (2) that BMD is related to age, BMI, quadriceps strength, and dietary calcium intake, where quadriceps strength is in turn determined by age and physical activity. These models were specified by a series of linear structural equations, using the SAS system's CALIS procedure. Parameters of these equations were estimated by the maximum likelihood method. All analyses were performed based on the variance-covariance matrix. Goodness-of-fit indices for the models were evaluated based on the chi-square statistic. A model that yields a low chi-square value (with high p value) is preferable to a model with high chi-square value (with low p value) for the same degrees of freedom, because the former indicates a better fit than the latter. Three additional goodness-of-fit indices were also considered, namely, the normed fit index,(38) the nonnormed index, and the comparative fit index.(39) The indices have values that range between 0 (for absolutely poor fit) and 1 (for a complete fit).


Subject characteristics

Data from 1075 women and 690 men whose BMD measurements were available were analyzed. The average and SD of age for both sexes was 69.5 ± 6.5 years old (Table 1), with an above-average concentration of subjects in the younger age group of 60–69 years (58%), compared with 70–79 years (33%) and 80+ years (9%). The BMI values in the sample were normally distributed for both sexes, with a mean of 26 ± 3.6 kg/m2 for men, almost identical to that in women (25.4 ± 4.6 kg/m2). Approximately one-third of women and 36% of men had a BMI value greater than 27 kg/cm2. Dietary calcium intake was skewed toward lower values; in both sexes, approximately 75% of calcium intake levels were less than 800 mg/day. The medians of intake for men (592 mg/day) and women (573 mg/day) were not significantly different. Quadriceps strength in men (33 ± 13 kg) was significantly higher (p < 0.0001) than that in women (20 ± 8 kg). The PAI value for men was also higher (35 ± 8.9; p < 0.001) compared with that for women (30 ± 4.4).

Advancing age was correlated with lower BMI (r = −0.18; p < 0.0001), quadriceps strength (r = −0.26; p < 0.0001) and PAI (r = −0.22; p < 0.0001) values in both men and women. A positive correlation between BMI and quadriceps strength values in women (r = 0.08; p = 0.007) was somewhat weaker than that in men (r = 0.13; p = 0.006). In contrast, the correlation between PAI and quadriceps strength values in women (r = 0.23; p < 0.0001) was higher than that in men (r = 0.11; p = 0.003). No significant correlation between dietary calcium intake (in logarithmic scale) and BMI or quadriceps strength values was observed in either men or women.

Table Table 1.. Characteristics of Study Subjects
 MalesFemalesp Value
  1. aIn natural logarithmic scale.

Number of subjects1075690 
Age (yr)69.1 ± 6.269.4 ± 7.00.302
Weight (kg)78.3 ± 12.665.2 ± 12.5<0.001
Height (cm)173.4 ± 6.9160.0 ± 6.3<0.001
Body mass index (kg/m2)26.0 ± 3.625.4 ± 4.60.008
Quadriceps strength (kg)33.2 ± 13.419.1 ± 8.0<0.001
Dietary calcium intake mg/day636 ± 338642 ± 3530.767
ln (mg/day)a6.3 ± 0.66.3 ± 0.60.905
Physical activity index35.2 ± 8.930.8 ± 4.4<0.001
Lumbar spine BMD (g/cm2)1.24 ± 0.201.03 ± 0.19<0.001
Femoral neck BMD (g/cm2)0.92 ± 0.140.79 ± 0.13<0.001

Associations between BMD versus age and BMI values

Lumbar spine BMD was positively associated with both age and BMI values in men by the following equation: Lumbar spine BMD = 0.0014 × Age + 0.016 × BMI + 0.722, R2 = 0.20. In women, it was positively associated with BMI values and negatively correlated with age, with the following regression equation: Lumbar spine BMD = −0.0032 × Age + 0.0133 × BMI + 0.912, R2 = 0.18. At the femoral neck site, for both men and women, BMD was positively associated with BMI values and inversely associated with age; for men, Femoral neck BMD = −0.004 × Age + 0.013 × BMI + 0.87 (R2 = 0.14), and for women, Femoral neck BMD = −0.007 × Age + 0.011 × BMI + 0.99 (R2 = 0.11).

The association between BMI and BMD values was predominantly linear, because higher-order terms such as quadratic equations did not improve the linear fit. Compared with subjects whose BMI was <27 kg/m2, those with BMI values >27 kg/m2 had higher age-adjusted BMD values (8% in men and 10% in women). Subjects whose BMI value was >30 kg/m2 (9.6% in men and 12% in women) had 10% (men) and 13% (women) higher age-adjusted femoral neck or lumbar spine BMD values than those subjects whose BMI value was less than or equal to 30 kg/m2. This difference was consistent for any given age group (data not shown).

Associations among dietary calcium intake, quadriceps strength, PAI, and BMD values

In women, dietary calcium intake was significantly associated with femoral neck, but not lumbar spine, BMD values, accounting for less than 1% of femoral neck BMD variance. Femoral neck BMD values among women with the highest dietary calcium intake tertile (>710 mg/day) were 2.5% higher than values among those in the lowest tertile (<460 mg/day).

Quadriceps strength was significantly associated with both lumbar spine and femoral neck BMD values. Variation in quadriceps strength accounted for approximately 4% of the variance in total lumbar spine and femoral neck BMD values, even after adjusting for the effects of age, BMI, and dietary calcium intake (Table 2).

In men, dietary calcium intake had a positive relationship with both lumbar spine and femoral neck BMD. After adjusting for the effects of age, BMI, or quadriceps strength and physical activity, the association remained statistically significant (Table 2). Femoral neck and lumbar spine BMD values among subjects in the highest dietary calcium intake tertile (>710 mg/day) were 5% higher than among those in the lowest tertile (<460 mg/day). However, variation in dietary calcium intake accounted for only 1% of the total variance of BMD.

The PAI was positively associated with femoral neck BMD in men, and with both femoral neck and lumbar spine BMD in women. However, after adjustment for age, BMI, and other factors in the multiple regression model, the association became statistically nonsignificant (Table 2).

Quadriceps strength was positively associated with femoral neck, but not lumbar spine, BMD values in men, accounting for approximately 3% of the total femoral neck BMD variance. The strength of association remained significant, after adjusting for age, BMI, and other factors (Table 2). BMD values among men with quadriceps strength of greater than or equal to 40 kg (top tertile) were 3% higher than those among men with the lowest tertile of quadriceps strength (≤28 kg).

Further analysis indicated a significant interaction between dietary calcium intake and BMI (p = 0.08 for males and p = 0.11 for females), dietary calcium intake, and quadriceps strength (p = 0.04 for males and p = 0.05 for females). This first interaction was characterized by the nonheterogeneity of the effects of dietary calcium intake on BMD between BMI tertiles. For example, in men, higher dietary calcium intake was associated with higher BMD only among subjects with BMI values ≤27 kg/m2, and not among heavier subjects (BMI values >27 kg/m2); similarly, in women, the effect was observed among thinner subjects (BMI values ≤23 kg/m2), and not among subjects with BMI values >23 kg/m2 (Fig. 1). On the other hand, the effect of quadriceps strength was significant among men with low, and women with high, dietary calcium intake (Table 3).

Risk of osteoporosis

By the WHO classification, 26% of women and 10% of men in this sample were “osteoporotic.” The prevalence increased with advancing age, from 14% among women aged 60 to 69 years to 37% among those aged 70 to 79 years and 61% among those aged 80+ years. For men, the corresponding figures were 6%, 14%, and 24%.

Table Table 2.. Dietary Calcium Intake, Physical Activity, and Quadriceps Strength and Bone Mineral Density
 Dietary calcium intake (ln mg/day)Quadriceps strength (kg)PAI (METS score)
  1. Values are the regression coefficients of relationship and standard error of estimate.

  2. aAge, BMI, dietary calcium intake, quadriceps strength, and PAI were all considered in the model.

  3. *p < 0.05, d̊p < 0.001.

  Lumbar spine BMD   
   Unadjusted analysis0.009 ± 0.0100.003 ± 0.0007d̊0.0019 ± 0.001*
   Adjusted for age0.015 ± 0.0100.002 ± 0.0007*0.0021 ± 0.001*
   Adjusted for age, BMI0.007 ± 0.0090.001 ± 0.0007*0.0004 ± 0.001
   Multiple regressiona0.006 ± 0.0100.001 ± 0.00070.0004 ± 0.001
  Femoral neck BMD   
   Unadjusted analysis0.019 ± 0.007*0.004 ± 0.0005d̊0.0032 ± 0.0009*
   Adjusted for age0.010 ± 0.006*0.001 ± 0.0005d̊0.0030 ± 0.0009*
   Adjusted for age, BMI0.016 ± 0.006*0.001 ± 0.0005d̊0.0001 ± 0.0008
   Multiple regressiona0.016 ± 0.006d̊0.001 ± 0.0004*0.0008 ± 0.0008
  Lumbar spine BMD   
   Unadjusted analysis0.043 ± 0.014d̊0.001 ± 0.00050.0004 ± 0.001
   Adjusted for age0.043 ± 0.012d̊0.001 ± 0.0010.0004 ± 0.001
   Adjusted for age, BMI0.042 ± 0.012d̊0.001 ± 0.0010.0001 ± 0.001
   Multiple regressiona0.043 ± 0.014d̊0.001 ± 0.0010.0001 ± 0.001
  Femoral neck BMD   
   Unadjusted analysis0.023 ± 0.010*0.003 ± 0.0005d̊0.0021 ± 0.001*
   Adjusted for age0.022 ± 0.001*0.002 ± 0.0005d̊0.0020 ± 0.001*
   Adjusted for age, BMI0.022 ± 0.001*0.001 ± 0.0005d̊0.0001 ± 0.0006
   Multiple regressiona0.021 ± 0.001*0.001 ± 0.0005d̊0.0001 ± 0.0006

The risk of osteoporosis was higher among subjects with lower dietary calcium intake, quadriceps weakness, and BMI values. For instance, men in the lowest tertile of dietary calcium intake were twice as likely to be osteoporotic than were their counterparts in the top tertile. A similar nonsignificant trend was observed in women. Men and women in the lowest tertile of quadriceps strength had 3.7-fold and 2.3-fold, respectively, greater risk of osteoporosis than did those in the highest tertile. Moreover, the risk of osteoporosis for those in the lowest tertile of BMI compared with those in the highest tertile of BMI was 4.6-fold (95% confidence interval [CI]: 3.4 to 6.0) greater for women and 3.3-fold (95% CI: 1.8 to 5.8) for men (Table 4).

Each of the three risk factors (BMI, quadriceps strength, and dietary calcium intake) was dichotomized into high risk (first tertile) versus low risk (second and third tertiles). For the three combined factors, there are eight possible combinations as shown in Table 5. In both sexes, the prevalence of osteoporosis was greater with the presence of two risk factors and greatest among the small proportion of subjects with all three risk factors. For example, in women, 31% of the population did not have any of the three risk factors, and among these subjects the prevalence of osteoporosis was 12%. Among women with two risk factors, the prevalence doubled, and among those who had all three risk factors (4.4% of the population), the prevalence of osteoporosis was 64%.

Similarly but to a lesser extent, in men the prevalence of osteoporosis was low (1.5%) among those without any risk factors (28.6% of the population). However, this prevalence rose by 5- to 10-fold among those with at least two risk factors and was highest among those with all three risk factors (40%) (Table 5).

Path analysis

Path analysis suggested that model II fitted the data adequately for femoral neck BMD (χ2 = 2.04, p = 0.36) in women and in men (χ2 = 4.63, p = 0.10). In both sexes, the goodness-of-fit index was 0.99. Model I did not fit the data for either men (χ2 = 30.21, p < 0.0001) or women (χ2 = 34.6, p < 0.0001). According to the estimated parameters of this model (Fig. 2A), age exerts its effect simultaneously on BMD and quadriceps strength but BMD is determined by quadriceps strength, BMI, and dietary calcium intake, with correlations between age and BMI as well. For females, the strongest determinant was BMI, accounting for 14% ([0.372]2 = 0.14) of the variance in BMD values, followed by age (10%). Quadriceps strength and dietary calcium intake together accounted for approximately 1% of BMD variance. Similar trends, but to a lesser extent, were observed in males (Figure 2B).

Figure Fig. 1..

Relationship among femoral neck BMD, BMI, and dietary calcium intake in (left panel) women and (right panel) men. In both sexes, the effect of dietary calcium intake was evident mainly among lean subjects (those with low BMI values). For women, the tertiles of BMI values are ≤15, 16–22, and ≥23 kg; for men, ≤28, 29–38, and ≥39kg.


Like many other complex diseases that express themselves in later decades of life, osteoporosis is probably not a single entity, but rather an end state arising from a cumulation of various disturbances of the skeletal homeostasis. If the aim is to prevent rather than to treat osteoporosis, adequate understanding of the roles of potentially modifiable lifestyle factors such as nutrition and physical activity must be addressed. This is particularly relevant for the elderly, because this group has the highest risk of osteoporotic fractures. Although it is commonly thought that nutrition and physical activity are important factors in bone health,(9,40–44) it has not been easy to show the relationship in the general population. The present study suggests that dietary calcium intake has a favorable, albeit modest, effect on BMD in both elderly men and women. Furthermore, the relationship between effect of dietary calcium intake and BMD was mainly evident among lean subjects with BMI values <27 kg/m2. Despite their modest individual effects, these risk factors, when combined with BMI, can identify a relatively high proportion of osteoporotic subjects. Moreover, quadriceps strength was positively associated with BMD, predominantly in the weight-bearing site of the femoral neck.

Table Table 3.. Quadriceps Strength and Dietary Calcium Intake Interaction on Femoral Neck Bone Mineral Density
 Dietary calcium intake (mg/day)a
Quadriceps strength tertile≤465466–709≥710
  1. Values are adjusted femoral neck BMD and ± standard error (g/cm2) by tertile of quadriceps strength and dietary calcium intake.

  2. *p = 0.002, d̊p < 0.001 level, versus first tertile of quadriceps strength.

≤15 kg0.74 ± 0.0100.76 ± 0.0110.76 ± 0.011
16–22 kg0.78 ± 0.0100.79 ± 0.0110.80 ± 0.010
≥23 kg0.79 ± 0.0120.80 ± 0.0110.85 ± 0.012*
≤28 kg0.85 ± 0.0150.94 ± 0.0150.90 ± 0.015
29–38 kg0.92 ± 0.016d̊0.92 ± 0.0170.93 ± 0.016
≥39 kg0.91 ± 0.016d̊0.95 ± 0.0160.94 ± 0.016
Table Table 4.. Risk Factors for Osteoporosis in the Elderly
Risk factor tertileOR* (95% CI)OR* (95% CI)
  1. *Values are age-adjusted odds ratio (OR) and 95% CI, relative to highest calcium intake, quadriceps strength, and BMI tertile, respectively, with a value of 1. Significant ORs (p < 0.05) are boldfaced.

Dietary calcium intake  
Middle1.12 (0.87, 1.45)1.02 (0.97, 1.07)
Lowest1.24 (0.97, 1.59)2.06 (1.23, 3.47)
Quadriceps strength  
Middle1.48 (1.10, 2.01)1.19 (0.53, 2.63)
Lowest2.31 (1.76, 3.03)3.74 (2.10, 6.65)
Middle2.26 (1.59, 3.20)1.53 (0.76, 3.09)
Lowest4.55 (3.43, 6.02)3.26 (1.83, 5.81)

Although animal studies support a relationship between dietary calcium and BMD,(45) studies on community-living humans have been less convincing. Several studies have reported a positive relationship between dietary calcium intake and BMD in peri- and postmenopausal women(5,6,10,12,43,46–48) or men.(7,49) However, others found no significant effect in women(50–54) or in men.(55–56) There are many possible reasons for these discrepancies, including differences in population characteristics, genetic background, and experimental design, wide variations in the ranges of calcium intakes, and differences in BMD measurement methods and sites. Furthermore, current levels of physical activity and calcium intake levels may not reflect the long-term levels, confounding analyses. It is interesting that in the present study, the dietary calcium effect on the femoral neck among women, and on both the femoral neck and the lumbar spine among men, was only apparent among leaner subjects. The strength of the association between dietary calcium intake and BMD observed in this study is modest, before or after adjusting for age and BMI, and higher weight (or BMI) was associated with higher BMD irrespective of recorded dietary calcium intake level.

Table Table 5.. Prevalence of Osteoporosis for Various Risk Factors
LeannessaQuadriceps weaknessbLow calcium intakecPopulation prevalenced(%)Percentage with osteoporosis
  1. Risk value was defined as relative to being in the first tertile of the risk factor or not.

  2. aFor BMI, the risk value was ≤24 kg/m2 in men and ≤23 kg/m2 in women.

  3. bFor quadriceps strength the risk value was ≤28 kg in men and ≤15 kg in women.

  4. cFor dietary calcium intake <465 mg/day in men and women.

  5. dPrevalence is the proportion of subjects with a combination of risk factors; the denominator of this estimate was 1075 women and 690 men.

Overall  100.025.8
Overall  100.010.0

Mechanical loading through physical activity is another potentially important factor in the preservation of bone mass in the elderly. Although even a short period of immobilization can result in significant bone loss,(18) whether exercise can improve or maintain bone mass in the elderly has not been clear. In this study, higher physical activity levels and higher quadriceps strength were associated with high BMD at the weight-bearing site of the femoral neck, but not at the lumbar spine, in both men and women. This finding is consistent with increasing activity improving or maintaining bone mass.(57) It is also consistent with other cross-sectional studies, in which high levels of physical activity are associated with greater bone density of the phalanx, lumbar spine, metacarpals, radius, ulna, and os calcis.(58–59) Associations between physical activity and BMD have not generally been adjusted for weight or height.(6,21) On the other hand, in several studies, there were no significant correlations between BMD and the physical activity level assessed by questionnaire, but there was a positive effect when bone was related to other measures such as work physical capacity,(60) total body potassium, and maximal heart rates.(61) Nonsignificant associations between muscle strength and BMD have also been reported previously.(62) However, although Sinaki et al.(63) reported that lumbar spine BMD and the strength of the back extensors were correlated before but not after adjustment for age and height, Pocock et al.(60) found that fitness was related to femur and lumbar spine BMD, as was weight. Some of these conflicting results may also relate to inadequate sample size. For example, to have an 80% chance of detecting a difference of 0.05 g/cm2 in femoral neck BMD between the top and bottom tertile of quadriceps strength (or dietary calcium intake), at least 600 subjects are required. Similarly, to reliably demonstrate a correlation of 0.10 between the PAI (or any traits) and BMD, a sample size of at least 1020 subjects is required.

These discrepancies also suggest that any effect of physical activity on bone is heterogeneous. For example, the effect of quadriceps strength on bone density was only apparent among women with dietary calcium intakes of >710 mg/day. Any effect of dietary calcium or physical activity is not necessarily linear and should not be considered in isolation from other environmental factors. Genetic factors need to be considered, because they may account for as much as 40% of variance of muscle strength,(64) and some of these genetic factors may have pleiotropic effects on bone density(65) (e.g., the vitamin D receptor gene(66)). Thus, both genetic and environmental factors likely contribute to the effects of nutrition and physical activity on bone health.

A positive association between BMI (or body weight) and bone density has been documented in a number of large-scale epidemiological studies.(35,67–68) Cross-sectional analyses from post- and premenopausal women suggest in some studies(69–70) that bone mineral level is related to fat mass and in other studies(65,68,71) to both lean and fat mass. The relationship may relate to how bone density is measured.(72) For example, fat mass is the main determinant of BMD when it is expressed in volumetric units (g/cm3); lean mass is the principal predictor of BMD when it is expressed in areal units (g/cm2).(73) Because muscle strength and BMD are all under strong genetic influence,(64–65,73) there appears to be a complex genetic and environmental interaction involving at least physical activity, dietary calcium, and genetic factors such as the vitamin D receptor gene,(74–76) the estrogen receptor, and the collagen I alpha 1 genes in the determination of BMD.

Although the dietary calcium and quadriceps strength effects were each relatively small (accounting for less than 2% of age- and BMI-adjusted variance of BMD), their combination with BMI can identify more subjects with bone density low enough to meet the diagnostic criteria of osteoporosis. For example, a little more than one-third of women and approximately 9% of men with low BMI had BMD-based osteoporosis. Those with either quadriceps weakness or low dietary calcium intake, in addition to low BMI, had a prevalence of osteoporosis of 51% in women and 14% in men. When all three factors (BMI, quadriceps strength, and dietary calcium) were considered, the prevalence increased to 60% in women and 40% in men.

Figure Fig. 2..

Path models of interrelationship among age, quadriceps strength, dietary calcium intake, BMI, and femoral neck BMD for (A) women and (B) men. The figure on each straight line is the standardized regression coefficient (i.e., the squared coefficient represents the proportion of variance of BMD that is explained by a risk factor). The figure on the curved line is the correlation coefficient (reflects the covariance between the two variables).

In this population, there was no significant association between lumbar spine BMD and quadriceps strength or physical activity in men, either before or after adjusting for age and BMI. In fact, lumbar spine BMD in men was found to increase slightly with advancing age. This is probably caused by the presence of osteoarthritis, which can artificially elevate lumbar spine BMD measurements in elderly men. Indeed, the higher degree of osteophytosis in men was associated with higher and significant increase in lumbar spine BMD. Subjects with any degree of osteophytosis had higher BMD values (21%) than did those without osteophytosis.(77) However, no such effect was observed in femoral neck BMD.

There are a number of limitations to generalizing the present findings. In a cross-sectional study, correlations cannot be taken as definitive evidence of causal relationships. Measurements of dietary calcium intake and PAI at a single time point may not reflect long-term exposure. Also, the qualitative assessments could include measurement errors and hence underestimate any true association between these factors and BMD. Protein intake may increase calcium loss and thus affect BMD in the elderly.(78) However, this effect was not examined in the present study, because the questionnaire does assess dietary protein intake. Finally, these results from subjects of white background may not be generalizable to other populations. Despite these potential limitations, the study was based on a large sample size and a homogeneous, unselected population, so differences could reliably be detected, which would not be possible with smaller studies.

In summary, these data suggest a positive and interactive association of BMD with BMI, quadriceps strength, and dietary calcium intake, which could potentially translate into practical strategies for the prevention of osteoporosis in the elderly. The findings, together with other data, support the recommendation that regular lifelong physical activity, in combination with adequate nutrition for both calcium intake and weight maintenance, should be part of a healthy lifestyle to enhance musculoskeletal health, and possibly improve the quantity of life in the aged.


We gratefully acknowledge the expert assistance of Janet Watters and Angela Ferguson in interviewing, collecting data, and bone densitometry. We acknowledge the invaluable help of the staff of Dubbo Hospital, particularly B. Luton, C. Mitchell, M. Russell, and B. Ayrton. Supported in part by the Australian Institute of Health and the Australian Dairy Corporation, and in full by the National Health and Medical Research Council of Australia.