Journal Logo


Previous physical activity relates to bone mineral measures in young women


Author Information
Medicine & Science in Sports & Exercise: January 1996 - Volume 28 - Issue 1 - p 105-112
  • Free


More than 25 million people in the U.S. are affected by the debilitating disease of osteoporosis, which is characterized by decreased skeletal mass and increased susceptibility to bone fractures. Bone mass is inversely related to the incidence of fracture in Caucasian women over the age of 50 (16). Efforts to reverse osteoporosis once the disease has developed have not been successful; therefore, studies have focused on prevention of osteoporosis. Physical activity is thought to be one environmental factor that may have an impact on bone mass(4).

Some (3,7,12), but not all(9,11), cross-sectional studies have demonstrated positive relationships between bone mineral measures and exercise or physical activity in premenopausal women. On the other hand, a controlled, nonrandomized 12-month exercise study (7) and an 8-month randomized exercise trial (17) in young women showed modest increases in bone mineral measures with exercise, whereas a prospective study (10) showed no influence of physical activity.

Cross-sectional studies suggest that the type of exercise may also differentially influence bone mass. Several studies have shown that weight lifters had greater bone mineral content (BMC, 12) or bone mineral density (BMD, 6, 13) than swimmers or those participating in aerobic activity. However, Block et al.(2) showed that trabecular bone density was greatest in individuals participating in both aerobic and weight-bearing activities compared with those participating in either type of activity alone.

Physical activity levels may contribute to modulating peak bone mass in young women. Understanding the age at which activity may influence specific bone sites, as well as the level and type of activity, will help investigators target specific treatments to increase bone mass and perhaps subsequently prevent osteoporosis.

Thus, the purpose of this study was to examine the relationship between previous physical activity and total body bone measures as well as bone measures at specific sites in young women. In addition, the relationship of type of previous activity (weightbearing vs non-weight-bearing) to these same bone mineral measures was examined. This study used a comprehensive list of physical activities developed for use in assessing the relationship of activity to various health outcomes and to improve the comparability of studies (1). These analyses give insights into the age, type, and level of activity that may have an effect on bone mineral measures in young women.



Females (N = 204, age 18-31 yr) were recruited through a variety of techniques including direct mail, radio, and flyer advertisements to participate in a study investigating the effects of an exercise intervention on bone strength. Participants were minimally active as defined by 2 h·wk-1 or less of exercise for the year before entry into the study. Exclusionary criteria included: intake of chronic medication that interferes with calcium absorption; fewer than nine menstrual cycles in the last year; history of high blood pressure, heart disease, diabetes, or malabsorption; and bone, kidney, or hormonal disorders that might affect calcium metabolism. The study protocol was approved by the Purdue University Institutional Review Board.

Previous Physical Activity Assessment

Previous physical activity was assessed by self-report questionnaire followed by an interview to improve accuracy (Appendix A). For consistency in collecting this information, all interviews were conducted by only two researchers. The questionnaire included information on all occupations held, but only the 5 yr before enrollment in the study were included in the analyses. Participants indicated whether they worked full or part time for each occupation reported. Since the exact number of hours worked per week was not obtained, for consistency in analysis it was assumed that full time equaled 40 h·wk-1 and part time equaled 20 h·wk-1. Years of participation in competitive sports were reported for both high school and college. Leisure activity was defined as all physical activity associated with seasonal (summer softball leagues for example) or otherwise occasional leisure pursuits, while exercise was defined as participation in an organized and regularly scheduled exercise program. In both cases, subjects estimated their participation in these activities during the last 5 yr including months per year, sessions per month, and minutes per session.

The Compendium of Physical Activity developed by Ainsworth et al.(1) was employed to estimate average daily energy expenditure (EE, kcal·d-1) for all categories of physical activity reported in the above questionnaire. The Compendium represents a standardized system with a comprehensive list of activities coded by function(leisure, competitive, or occupational), intensity (with respect to energy expenditure), and specific type of activity. The compendium list is expressed in METs (metabolic equivalents), and data from the questionnaire were used to generate results expressed as kcal·d-1. Energy expenditure estimates for occupational, leisure activity, and exercise over the last 5 yr were combined for analyses (occupation + leisure EE). Daily high school EE and college EE as a result of sport participation during those years were estimated by assuming 3 months participation for each sport season. The 3-month period was employed since most schools have fall, winter, and spring sport seasons and approximately a 9-month school calendar. It was further assumed that a minimum of 10 h·wk-1 participation would occur as a result of practices or competition lasting at least 2 h-d-1 and 5 d·wk-1. Although some athletes may participate to varying degrees in practices and competitions related to one sport year-round, the above assumptions were made to provide some consistency. Five-year EE was derived from 5-yr occupation + leisure EE + high school and/or college sports EE if high school and/or college were within the last 5 yr. All of the above coding was completed by one investigator, and energy expenditures reported represent those above resting EE and usual daily maintenance activity. Further, each activity was coded according to whether it was primarily non-weightbearing (swimming, biking, secretary) or weightbearing (volleyball, aerobics, waitress).

Fitness Assessment

Weight (kg) was measured with a regularly calibrated electronic scale and height (cm) was measured with a wall-mounted stadiometer with subjects in light clothing and without shoes. Current oral contraceptive use was determined during the interview.

Maximal oxygen uptake (˙VO2, ml·kg-1·min-1) was assessed with a walking test at 5.65 km·h-1 (3.5 mph) on a motorized treadmill. After a 3-min warm-up, elevation of the treadmill was increased by 2% every 2 min. Expired air was sampled (2900 C Metabolic Measurement System, SensorMedics, Anaheim, CA) and heart rate was measured (Polar Vantage SL, Polar USA, Stamford, CT) continuously. Criteria for termination included at least two of the following: volitional fatigue, <3% change in oxygen uptake with increased elevation, heart rate at or near age-predicted maximum heart rate, or respiratory exchange ratio > 1.0.

Bone Mass Measurements

Total body, spine (lumbar vertebrae 2-4) and femoral neck bone mineral density (g·cm-2), and bone mineral content (g) were assessed with a dual-energy absorptiometer (DPXL, Lunar Corp., Madison, WI). Subjects were wearing no metal when the scan was done. Short-term precision was determined by the standard deviation of two measurements repeated on the same day divided by the mean. Short-term precision for this instrument for total body, spine, and femoral neck was 0.8%, 1.04%, and 1.4%, respectively. The radial bone mineral density (g·cm-2) and content(g·cm-1) were performed using a single photon absorptiometer(SPA, Lunar SP2, Lunar Corp.). The short-term precision of SPA of the midshaft radius is 1.6%.

Statistical Analysis

Data were analyzed for normality and homogeneity of variance. Analyzing the data converted to log did not effect the results and thus the data were left unconverted to simplify the reading of the manuscript. Pearson correlation analyses were performed with all variables and partial correlations performed using selected variables. A series of stepwise multiple regressions were performed for each site to predict the amount of variance in the bone mineral measurements that could be explained by EE variables. Criteria for inclusion into the stepwise model was P ≤ 0.05. The dependent and independent variables as well as any covariates for each stepwise analysis are described in the results section.


Subject characteristics are described in Table 1(means ± SD) and measures of previous activity levels, expressed in kcal·d-1 in Table 2 (means ± SD). If no activity was reported in a category by an individual, zero values were assigned and data for all subjects were included in calculation of the means and SD for that category. Few subjects participated in college athletics(N = 32) and non-weightbearing high school athletics (N = 25). Due to the low level of participation in college athletics, this measure was not included separately in further analyses.

The correlations between previous activity measures(kcal·d-1), age, weight, ˙VO2max, and height are shown in Table 3. Age and height significantly correlated with all measures of previous activity except high school, weight correlated with all measures including high school, and ˙VO2max levels correlated with high school EE only.

Five-year EE was significantly correlated with all bone mineral measures(Table 4). The significance level of P < 0.05 as well as P < 0.0007, which is the appropriate Benferroni alpha-level adjustment for the number of correlations presented, are indicated in Table 4. Occupation + leisure EE correlated with all bone mineral measures except spine BMD. High school EE correlated with all bone mineral measures except radius BMD. ˙VO2max correlated only with femoral neck BMD. Results in all cases were similar when age, height, or current oral contraceptive use was controlled in the analyses (data not shown).

When controlled for weight, 5-yr EE and occupation + leisure EE remained positively associated with all BMC, but not BMD, measures(Table 4). High school EE remained associated with total body BMD and BMC, femoral neck BMD, and spine BMC.

When analyses were performed controlling for total body BMD, the significant correlations remained between 5-yr EE and occupation + leisure EE and all BMC measures (Table 4). When total body BMD was controlled, high school EE was significantly correlated with femoral neck BMD and total body BMC.

Tables 5 through 8 each contain the results of seven stepwise multiple regression analyses to predict the amount of variance in bone mineral measurements that can be explained by the various EE variables. The independent variables for Table 5 include 5-yr EE, occupation + leisure EE, and high school EE. Results indicated that 5-yr EE or occupation + leisure EE were significant predictors of all measures of BMC and BMD except the femoral neck BMD. In contrast, high school EE was a significant predictor of femoral neck BMD as well as total body BMD and BMC, and spine BMC.

Table 6 reports the predictive ability of the same independent variables as Table 5 when body weight is controlled. Body weight was added to the model as a covariate because weight is a factor in determining total kcal·d-1 and is itself a significant predictor (P < 0.05) for all bone measures. After controlling for weight, 5-yr EE or occupation + leisure EE predicted total body BMC, radius BMD, and BMC. High school EE was a significant predictor for total body BMD and BMC, femoral neck BMD and spine BMD and BMC.

For Table 7 the independent variables are non-weightbearing and weightbearing physical activity. Results demonstrated that non-weightbearing occupation + leisure EE was a significant predictor of all bone mineral measures except femoral neck BMD. Weightbearing occupation + leisure EE predicted radius BMD, total body BMC, and spine BMC. Weightbearing high school EE predicted all bone mineral measures except radius BMD.

Table 8 reports the predictive ability of non-weightbearing and weightbearing physical activity when weight was controlled. Non-weightbearing occupation + leisure EE was a predictor for total body BMC and spine BMC whereas weightbearing occupation + leisure EE was a predictor for radius BMD. In contrast, high school EE predicted all measures of bone health except radius BMD.


In this study, relationships between previous physical activity and bone mineral measures were determined. Activity was expressed in kcal·d-1 and by weightbearing or non-weightbearing activity. This allowed assessment of activity levels on a continuum with respect to energy expenditure as well as by type of activity that may stress the bones differently. The variance explained by the previous physical activity models on bone mineral measures ranged from total r2 = 0.04 to r2 = 0.18 in multiple regression analyses. Since weight has a strong influence on bone mineral measures, it is important to consider this factor in the analyses to determine the specific independent effects of physical activity. When weight was controlled, 5-yr EE or occupation + leisure EE remained significant predictors of total body and radius BMC. In addition, high school EE remained a significant predictor for all measures of bone health except radius BMD and BMC. Further, weightbearing and non-weightbearing activity both influenced various bone mineral measures. This suggests that physical activity itself, independent of weightbearing influence, is important to bone mineral measures.

Other cross-sectional studies have also demonstrated a relationship between physical activity and bone mineral measures(3,7,12). One cross-sectional analysis of life-time physical activity showed that 45 min of moderate to strenuous activity 4 times·wk-1 in 181 women aged 20-50 yr was associated with higher radius BMD and BMC (7). Similar results were shown in a study of 38 women aged 24-28 yr where activity that caused a sustained increase in heart rate for >90 min·wk-1 was associated with higher radial BMC and BMD (12).

Several intervention studies have examined the relationship between physical activity and bone health as well. One study (N = 72) assessed premenopausal women who were matched for age, BMI, and baseline activity. Participants volunteered to participate in either an exercise(weight lifting) or a control group. After 1 yr, a significant difference was observed between the groups in lumbar spine BMD, with a slight increase in the exercisers (0.81%) and a decrease in the control (0.5%)(5). For another controlled exercise trial, the participants were randomly assigned to an 8-month exercise program of resistance training, jogging, or a control group (17). The mean age was 19.9 + 0.7 yr and the final number in each group was 8(control), 10 (runner), and 12 (weight lifters). Lumbar BMD increased in both the runners (1.3%) and the weight lifters (1.2%), but no change was noted in the proximal femur. These studies support the results of the current study in that they demonstrate that participation in physical activity, even in the short-term, may impact on at least one bone mineral measure.

Generally, cross-sectional studies have shown that physical activity may explain between 6-20% of the variation in bone mineral measures(4). This range is consistent with the variance in bone mineral density explained by physical activity in the current study. When weight was considered, less of the variance (2.3-7.0%) in bone mineral measures was explained.

Below is an example of how one of the regression analysis reported inTable 7 can be employed to predict the effects of specific activities on a bone mineral measure at a specific site. The model employed in this study is the relationship of the average daily EE for the previous 5 yr, or total HS sport participation, to current bone mineral measures. The dependent variable (y) selected for this example is total body BMC (g). Thus, the multiple regression analysis at this site produced the following equation: Equation

This equation can be interpreted as follows: starting with 2279 g of total body BMC, for each kcal·d-1 increase in nonweightbearing occupation + leisure EE, BMC would increase 0.504 g; for each kcal·d-1 increase in weightbearing occupation + leisure EE, BMC would increase an additional 0.187 g; and for each kcal·d-1 of high school EE, BMC would increase an additional 0.788 g.

Employing the above equation, if a participant (63 kg) reported working full time (40 h·wk-1 and 50 wk·yr-1) as a secretary (METs = 1.5 kcal·kg-1·hr-1), her average daily EE would equal 518 kcal·d-1 ((1.5 METs·63 kg·40 hr·wk-1·50 wk·yr-1)/(365 d·yr-1)) as a result of her occupation. Entering this into the equation for non-weightbearing occupation and assuming all other factors are zero would result in an additional total body BMC of 261 g (0.504·518). From the intercept level of BMC (2279 g), this is an increase in total body BMC of 11.4% (261 g/2279 g). Similarly, kcal·d-1 of weightbearing occupation + leisure EE and high school EE could be added to the above equation. For example, if the secretary played 4 hr·wk-1, 12 wk·yr-1 of casual volleyball (METs = 3) it would result in an additional EE of 24.9 kcal·d-1 ((3 METs·63 kg·4 hr·wk-1·12 wk·yr-1)/(365 d·yr-1)), and total body BMC would increase an additional 0.2%(0.187·24.9 = 4.65; 4.65 g/2279 g = 0.2%). Moderate walking for 1 hr·d-1, 52 wk·yr-1, (METs = 3.5, 219.9 kcal·d-1) would add 1.8% (0.187·219.9 = 41.1; 41.1/2279 = 1.8%). If, on the other hand, the secretary ran briskly, 6 miles·hr(METs = 10.5) for 1 hr·d-1, 52 wk·yr-1, (628.3 kcal·d-1), total body BMC would increase an additional 5.16%(0.187·628.3 = 117.5; 117.5/2279 = 5.16%). Lastly, if the secretary participated in high school sports such as tennis (10 hr·wk-1; 12 wk·yr-1), following the same reasoning (METs = 7; 145 kcal·d-1), total body BMC would increase 5% from the intercept(0.788·145 = 114.0; 114.0/2279 = 5.0%). Thus, both hours and intensity contribute to the influence of activity on bone mineral measures. Though some of these predicted increases are small, they may represent a clinically important protection against osteoporosis since vertebral fractures are inversely proportional to bone mineral content in Caucasian women over the age of 50 yr (16). Variability not accounted for by these models might be explained by such factors as genetic predisposition, nutritional intake and other lifestyle health habits.

In the current study, ˙VO2max correlated with high school EE, and both of these factors correlated with femoral neck BMD. These results are consistent with the results of Pocock et al. (14), who demonstrated that fitness was a significant predictor of femoral neck BMD. High school EE results are supported by Theintz et al.(18), who found that femoral neck BMD does not increase beyond the age of 16 yr. Thus, it may be that participation in activity during high school has more impact on the femoral neck BMD than other bone sites.

A bias may exist in analyses of cross-sectional studies since those who participated in athletics may have had higher levels of BMC or BMD initially. Though this possibility cannot be eliminated in a cross-sectional analysis, the current data were analyzed controlling for total body BMD to minimize the potential that higher bone density, generally, may encourage participation in activities. This strategy assumes that one could have exercise-induced site-specific changes without great impact on total body bone mineral density. When total body BMD is controlled, the association of femoral neck BMD and total body BMC with high school EE, and activity in the 5 yr with BMC of total body, spine, and radius remained, further supporting the importance of these observations. Nevertheless, cross-sectional analyses must be carefully interpreted and the results used to direct future longitudinal studies.

The collection of the physical activity data may have been limited by the questionnaire employed as well as the subjects' ability for accurate long-term recall. The questionnaire was a modification of an instrument previously employed in a similar study (6). However, to increase the completeness and accuracy, the self-report assessment was followed with an interview by one of two trained researchers. Also, results obtained with the questionnaire employed in this study are similar to results obtained with similar instruments. However, continued examination of all available physical assessment tools is warranted.

These results suggest that high school athletics was a significant predictor of femoral neck BMD and that higher activity levels within this age range may also improve total body and spine BMC, as well as total body BMD. Thus, high school athletic participation and increases in occupation + leisure EE in young women may increase peak bone mass, and reduce the risk of osteoporosis.


1. Ainsworth, B. E., W. L. Haskell, A. S. Leon, et al. Compendium of physical activities: classification of energy costs for human physical activity. Med. Sci. Sports Exerc. 25:71-80, 1993.
2. Block, J. E., H. K. Genant, and D. Black. Greater vertebral bone mineral mass in exercising young men. Clin. Invest. 145:39-42, 1986.
3. Davee, A. M., C. J. Rosen, and R. A. Adler. Exercise patterns and trabecular bone density in college women. J. Bone Miner. Res. 5:245-260, 1990.
4. Forwood, M. R. and D. B. Burr. Physical activity and bone mass: exercises in futility? Bone Miner. 21:89-112, 1993.
5. Gleeson, P. B., E. J. Protas, A. D. Leblanc, V. S. Schneider, and H. J. Evans. Effects of weight lifting on bone mineral density in premenopausal women. J. Bone Miner. Res. 5:153-158, 1990.
6. Going, S., T. Lohman, R. Pamenter, et al. The effects of weight training on regional bone mineral density (BMD) in premeno pausal females. Med. Sci. Sports Exerc. 23(Suppl. 4):S115, 1990.
7. Halioua, L. and J. J. B. Anderson. Lifetime calcium intake and physical activity habits: independent and combined effects on the radial bone of healthy premenopausal Caucasian women. Am. J. Clin. Nutr. 48:534-541, 1989.
8. Henrich, C. H., S. B. Going, R. W. Pamenter, C. D. Perry, T. W. Boyden, and T. G. Lohman. Bone mineral content of cyclically menstruating female resistance and endurance trained athletes. Med. Sci. Sports Exerc. 22:558-563, 1990.
9. Lloyd, T., J. R. Buchanan, S. Bitzer, C. J. Waldman, C. Myers, and B. G. Ford. Interrelationships of diet, athletic activity, menstrual status, and bone density in collegiate women. Am. J. Clin. Nutr. 46:681-684, 1987.
10. Mazess, R. B. and H. S. Barden. Bone density in premenopausal women: effects of age, dietary intake, physical activity, smoking, and birth-control pills. Am. J. Clin. Nutr. 53:132-142, 1991.
11. McCulloch, R. G., D. A. Bailey, S. Houston, and B. L. Dodd. Effects of physical activity, dietary calcium intake and selected lifestyle factors on bone density in young women. Can. Med. Assoc. J. 142:221-227, 1990.
12. Metz, J. A., J. J. B. Anderson, and P. N. Gallagher. Intakes of calcium, phosphorus, and protein, and physical-activity level are related to radial bone mass in young adult women. Am. J. Clin. Nutr. 58:537-542, 1993.
13. Nilsson, B. E. and N. E. Westlin. Bone density in athletes. Clin. Orthop. 77:179-182, 1971.
14. Pocock, N. A., J. A. Eisman, M. G. Yeates, P. N. Sambrook, and S. Eberl. Physical fitness is a major determinant of femoral neck and lumbar spine bone mineral density. J. Clin. Invest. 78:618-621, 1986.
15. Recker, R. R., K. M. Davies, S. M. Hinders, R. P. Heaney, M. R. Stegman, and D. B. Kimmel. Bone gain in young adult women.J. A. M. A. 268:2403-2408, 1992.
16. Smith D. M., M. R. A. Khairi, and C. C. Johnston. The loss of bone mineral with aging and its relationship to risk fracture.J. Clin. Invest. 56:311-314, 1975.
17. Snow-Harter, C., M. L. Bouxsein, B. T. Lewis, D. R. Carter, and R. Marcus. Effects of resistance and endurance exercise on bone mineral status of young women: a randomized exercise intervention trial.J. Bone Miner. Res. 7:761-769, 1992.
18. Theintz, G., B. Bichs, R. Rizzoli, et al. Longitudinal monitoring of bone mass accumulation in health adolescents: evidence for a marked reduction after 16 years of age at the levels of lumbar spine and femoral neck in female subjects. J. Clin. Endocrinol. Metab. 75:1060-1065, 1992.

Appendix A





©1996The American College of Sports Medicine