Journal Logo

CLINICAL SCIENCES: Clinically Relevant

Impact of exercise on bone health and contraindication of oral contraceptive use in young women


Author Information
Medicine and Science in Sports and Exercise: June 2001 - Volume 33 - Issue 6 - p 873-880
  • Free


Physical activity has been associated with the attainment and maintenance of optimal peak bone mass, a determinant of risk for osteoporosis (6,10). Both cross-sectional and longitudinal studies show that increased physical activity increases skeletal mass during growth (2,28,31). Less is known about the effect of physical activity on skeletal mass in women during consolidation of bone mass. On average, by age 18, almost 92% of total body (TB) bone mineral content (BMC) has been attained, and by age 26, 99% has been attained (33). However, age of attainment of peak bone mass is site specific. Peak bone mass appears to be complete by age 16 in the femoral neck (34), whereas in the lumbar spine, bone mass increases throughout the third decade of life (22). Thus, the effects of exercise on specific skeletal sites may be age dependent.

Observational studies have shown the importance of physical activity on increasing bone mass in young women (22,32,37). Only a few randomized exercise intervention trials have been reported in premenopausal women with results ranging from a loss of 4%(25), no effect (27), to increases in bone mineral density (BMD) on the order of 1–3%(3,11,17,29). These studies varied with respect to the inclusion of women acquiring peak bone mass. Only the randomized trials of Snow-Harter et al. (29), Friedlander et al. (11), and Bassey et al. (3) included higher-impact exercises, and none addressed the effect of oral contraceptive (OCont) use.

Oral contraceptive use is a common practice among this age group. Because OCont use has been associated with reduction of endogenous sex steroids (4) and estrogen suppresses bone remodelling (21), OCont use could modify the effects of exercise on bone. Thus, the purpose of this study was to determine the effect of quantified resistance plus high-impact exercise training on BMD or BMC as modified by age and oral conceptive use in a randomized intervention trial in healthy, nonathletic women, aged 18–31 yr.


Subjects and protocol.

Women (N = 179, 18–31 yr) were recruited through a variety of techniques. Participants were minimally active (2 h·wk-1 or less of exercise) for the year before the study as assessed by self-report questionnaire followed by an interview to improve accuracy (32). Exclusion criteria included: intake of chronic medication or bone, kidney, or hormonal disorders that might affect calcium metabolism; less than nine menstrual cycles in the year before the study; and history of high blood pressure, heart disease, diabetes, or malabsorption. The women had not been pregnant or lactating for the prior 3 months. All subjects using OConts were taking low-dose estrogen compounds (50 μg or less ethinyl estradiol). When eligible subjects completed baseline testing as described below, they were categorized by age (18–23 and 24–31 yr) and OCont use, resulting in four groups. They were then randomized into either the exercise group (N = 107) or the nonexercise group (N = 72). Because of higher expected attrition in the exercise group, 50% more subjects were randomized into this group. After randomization, 141 subjects initiated the protocols of the study. Reasons for withdrawal were time constraints (31%), pregnancy (11%), moving (4%), medical (4%), personal (2%), and end of funding period (9%). Attempts were made to contact all subjects who withdrew from the study to obtain 24-month data. However, only two subjects who withdrew returned for 24-month data collection. The study protocol was approved by the Purdue University and Indiana University Institutional Review Boards, and all subjects provided written, informed consent.

Exercise protocol.

The recreational sports complex at Purdue University includes a Universal super circuit room, where the resistance exercise portion (Universal weight machines, Cedar Rapids, IA) of the protocol was performed. The super circuit consisted of two sets of eight upper body and eight lower body weight stations with a cycle ergometer located between each station. The overall exercise protocol for this study was as follows: 1) three sessions/wk of the super circuit described above; and 2) 60 min·wk-1 of jumping rope. All participants were oriented to the super circuit by one of two research assistants. During the orientation they were instructed on proper use of each weight station, which included how to log each visit on standard forms that were kept at the site. Once oriented, participants were free to attend the facility at any time during regular hours (58 h·wk-1). The super circuit is not supervised, but an audiotape runs continuously and provides instructions on when to start and stop exercising at each station and move on to the next station (30 s per exercise bout and 15 s to change stations and adjust machines/weight stacks/ergometers as appropriate). After 4 wk of adaptation, strength was assessed (one-repetition maximum tests) at each weight station in order to prescribe exercise at 70% of maximum strength. One-repetition maximum tests were repeated every 6 months thereafter to increase exercise prescriptions and to assess improvements in strength. Participants were also instructed to perform 8–12 repetitions at each station and to increase weight on their own in between assessments when they felt that they could do more than 12 repetitions if time allowed. Rope jumping was performed either at the exercise facility in a multi-purpose room or at home, and there were no restrictions placed on form or minimum number of min required in each bout of jumping. The only requirement was to accumulate 60 min of jumping each week. The quantitative forces and stresses of this exercise program on the femoral neck peak strain were quantitated using a model including anthropometric measurements, kinematic data, and kinetic data as previously reported (1). Kinematic data for jumping rope and walking were obtained using two-dimensional sagittal plane motion analysis, and kinetic data for these activities were obtained from force plate analysis. Peak strain rates were of the same order of magnitude for all resistance exercises (0.0005–0.0035-1). Peak strain rates of 0.028–0.055 s-1 for walking (the major component of physical activity not associated with the intervention) and jumping rope were significantly higher than for resistance exercises. Jumping rope to ∼250 mm for the total body center of gravity (140 mm for toe height) produced average ground reaction forces of 1.9 body weight (range 1.28–2.45) and average joint contact force at the hip of 3.14 BW (range 2.1–5.5 BW) on one leg.

Compliance was assessed at 6-month intervals by changes in lean mass (kg, DXA, Lunar Corp., version 1.2, Madison WI), changes in strength (lb), and on-site records of the weight-lifting sessions maintained by participants. In addition, subjects kept daily records that included physical activity, minutes jumping rope, and menstrual cycle information. Fitness (maximal oxygen uptake, V̇O2 mL·kg-1·min-1) was assessed with a walking test on a motorized treadmill (32). Incentives and motivational efforts included remuneration, prizes, frequent contact by investigators, and group activities.

Lifestyle factors.

Weight in light clothing was measured with a calibrated electronic scale, and height without shoes was measured with a wall-mounted stadiometer. Dietary intake of calcium was determined with a food frequency questionnaire (5) interview with additional questions on menstrual status and OCont use. For consistency in collecting this information, all interviews were conducted by only two researchers.

Bone measurements.

Spine (lumbar vertebrae 2–4), hip (femoral neck, greater trochanter, and Ward’s area) BMD and TB and spine BMC were measured with dual DXA energy x-ray absorptiometry (DXA, Lunar Corp.). The radial BMD and BMC were measured using a single photon absorptiometer (SPA, Lunar SP2, Lunar Corp.). Short-term precision was determined by the standard deviation of two measurements repeated on the same day and divided by the mean, and long-term precision was determined by the square root of the mean standard error divided by the overall mean. The short-term precision for spine BMD, radius BMD, and femoral neck BMD was 1%; and for TBBMC, spine BMC, radius BMC, greater trochanter BMD, and Ward’s area BMD was less than or equal to 2%. Long-term precision was as follows: TBBMC (2.2%, spine BMC (2%), spine BMD (1.2%), radius BMC (2.3%), radius BMD (1.5%), femoral neck BMD (2.3%), trochanter BMD 2.5%, Ward’s area BMD (3.3%). Data are expressed as spine, radius, femoral neck, greater trochanter, and Ward’s area BMD (g·cm-2) and TB (g), spine (g), and radius BMC (g·cm-2).

Biochemical markers of bone remodeling.

Biochemical markers for bone formation (serum osteocalcin and alkaline phosphatase) and resorption (serum tartrate resistant acid phosphatase and urinary hydroxyproline/creatinine) were assessed at baseline and 6-month intervals (36). Osteocalcin was measured by radioimmunoassay (RIA) using a rabbit polyclonal antiserum raised against bovine osteocalcin, a bovine osteocalcin as standard, and radioiodinated bovine osteocalcin as label. The interassay coefficient of variation (CV) was 9% at a serum concentration of 16 ng·mL-1. Total serum alkaline phosphatase was determined enzymatically by standard techniques using paranitrophenol as substrate. The interassay CV was 5%. Serum tartrate alkaline phosphatase was determined enzymatically by standard techniques using paranitrophenolphosphates as substrate. Serum was incubated for 1 h at 36°C before assay to destroy inhibitors. The interassay CV was 12% at a serum concentration of 10.8 U·L-1. After acid hydrolysis of urine, hydroxyproline was derivatized with phenylisothiocyanate and quantitated by ultraviolet absorbance after high-performance liquid chromatography (HPLC) separation using a Waters PICo Taq column eluted with a sodium acetate buffer. The interassay CV was 9.6% at a urinary concentration of 1.6 mg·mL-1. Urine creatinine (CR) was measured by a Roche multichannel analyzer (Roche, Somerville, NJ) by standard methods and expressed as the hydroxyproline/creatinine milligram ratio. Serum and second void urine were collected after fasting between days 3–12 of the follicular phase of the menstrual cycle.

Statistical analyses.

The primary outcome variables were analyzed using a three-way analysis of variance with independent factors age, exercise, and OCont use, each at two levels. Multivariate analysis of variance was used to compare the profiles of the subjects who dropped out with those who did not. A log transformation was performed to reduce the skewness of the serum and urinary measure. All computations were performed using SAS statistical software (26). Other analytical strategies including repeated measures of those completing all four time points (N = 52) showed similar results to all those described.


Characteristics of subjects at baseline for the four groups formed by exercise and OCont use are shown in Table 1. A three-way MANOVA with factors age group (18–23 and 24–31 yr), OCont use, and exercise group for these 11 baseline variables (age is excluded) indicated no significant differences with the exception of a main effect of age. Compared with subjects aged 18–23 yr, those who were 24–31 yr were heavier (64.1 vs 60.2 kg), taller (166.8 vs 163.6 cm), had greater TBBMC (2635.7 vs 2474.1 g), greater spine BMC (51.9 vs 48.6 g), greater radius BMC (0.88 vs 0.85 g), and less Ward’s BMD (0.95 vs 1.01 g·cm-2).

Table 1:
Subject characteristics at baseline (N = 141).

A total of 99, 84, 67, and 55 subjects completed 6, 12, 18, and 24 months, respectively. A comparison of subject characteristics among those who completed the study with those who did not indicated no significant difference (Wilks’ lambda = 0.74;F = 1.06;df = 29,89;P = 0.40; adjusted for age group and OCont use). The dropout rate for OCont users was not significantly different from the rate for nonusers (chi square = 0.39, df = 1, P = 0.53); similarly the dropout rates for the exercise groups were indistinguishable (chi square = 0.50, df = 1, P = 0.48).

At 24 months, reported exercise prescription completion was 46.7 ± 4.0 and 43.7 ± 6.3% for weight lifting and jumping rope, respectively (Table 2). OCont users had a higher compliance rate than nonusers for the four time periods. No changes in menstrual cycle were noted in any of the subjects. Although this protocol was not expected to impact aerobic fitness levels, the exercisers had an increase and the nonexercisers had a decrease in fitness (change in maximal oxygen uptake) that reached significance between the groups at 12 months (P = 0.02). Adherence to the protocol was evident by increases in strength (Table 2) and increases in lean mass from baseline (Fig. 1). Because differences due to OC use were not observed, strength, and lean mass data were combined by exercise assignment.

Table 2:
Compliance to exercise protocols.
Increases in lean body mass after exercise intervention. Changes from baseline that were significantly different from zero are denoted by an asterisk (*). Different letters denote a significant difference between the groups. Data are expressed as mean ± SE. The number of subjects at each time point is indicated below the point.

Exercise resulted in a 1–2% increase in TBBMC whereas TBBMC decreased by 1–2% in nonexercisers (Fig. 2). On the other hand, the femoral neck BMD was reduced (P ≤ 0.05) in the exercise group at 6 and 24 months compared with the nonexercise group (Table 3). For TBBMC, radius BMD, radius BMC, femoral neck BMD, and trochanter BMC and Ward’s BMD, there were no interactions among exercise group, OCont (reason for collapsing by exercise group) use, and age group. Independent of exercise and age, change in TBBMC from baseline was negative in OCont users (−0.91 ± 0.46%) and positive in nonusers (1.11 ± 0.54%) at 24 months (P = 0.007). Independent of exercise and OCont use, the younger women had a positive (0.92 ± 0.52%) change in TBBMC compared to the older women (−0.72 ± 0.46%, P = 0.007). No other bone sites had significant exercise effects except for spine BMC and BMD (Table 3 and below). In summary, exercise had a positive impact and OCont use had a negative impact on TBBMC, regardless of age. However, even within this age range, women who were not exercising lost TBBMC (Fig. 2).

Percent change in TBBMC after exercise intervention. Changes from baseline that were significantly different from zero are denoted by an asterisk (*). Different letters denote a significant difference between the groups. Data are expressed as mean ± SE. The number of subjects at each time point is indicated below the point. Results were similar using repeated measures including only those subjects who completed all four time points.
Table 3:
Percent change in bone mineral measures (Mean ± SE)

For spine BMC, there was a significant interaction between exercise and OCont use (Fig. 3). The exercise plus OCont group was not different in any parameter measured at baseline from the other groups, yet exercise plus OCont use resulted in a decrease at 6 months which returned to baseline by 1 yr. The other three groups showed a pattern of increase over time. Despite a higher compliance rate for the OCont users, spine BMC was less in the exercise OCont group compared with both the exercise no Cont group and nonexercise OCont group. Spine BMC decreased in 20 of the 27 (74%) OCont users in the exercise group at 6 months (sign test, P = 0.02) and was accompanied by a decrease in total serum calcium (−0.25 ± 0.69 mg·dL-1, P = 0.08). The results for spine BMD were similar. Thus, exercise in combination with OCont use compromised spine BMC and BMD.

Effect of exercise on percent change in spine BMC in oral contraceptive (OCont) users and non-OCont users after exercise intervention. Changes from baseline that were significantly different from zero are denoted by an asterisk (*). Different letters denote a significant difference between the groups. Data are expressed as mean ± SE. The number of subjects at each time point is indicated below the point. Results were similar using repeated measures including only those subjects who completed all four time points.

The relationship of serum and urine markers for bone remodeling (osteocalcin, tartrate resistant acid phosphatase, hydroxyproline/creatinine, and alkaline phosphatase), urinary calcium/creatinine, serum calcium, calcium intake, fitness, fat mass, lean mass, percent fat, weight, height, and postmenarchal age on the changes in bone reported above were examined. Although at 6 months percent fat was positively associated with the percent change in spine BMD (P = 0.04), no important differences in the results were found when it was used as a covariate.

No consistent patterns of change over time were evident in the biochemical markers for bone turnover, which is not surprising as there were both positive and negative bone changes occurring. At baseline, four markers of bone turnover were lower for OCont users (Table 4). A two-way MANOVA with independent factors OCont use and exercise group for the five variables indicated a significant main effect for OCont use (Wilkes lambda = 0.87, F = 3.55, df = 5,122, P = 0.005), but the main effect of exercise and the interaction were not significant (P = 0.28 and P = 0.39, respectively).

Table 4:
Biochemical markers of bone turnover at baseline (N = 141).


This study demonstrated that exercise resulted in a positive effect on TBBMC, which was greater in the younger than older women. On the other hand, exercise had a negative effect on femoral neck BMD. An unexpected interaction between exercise and OCont use was observed such that spine BMC and BMD decreased during the first 6 months and remained lower for the 2-yr intervention.

The percent change after 24 months in TBBMC for those who participated in the program of resistance and high-impact moderate exercise increased and the percent change in TBBMC of nonexercisers decreased. The loss of TBBMC in premenopausal nonexercisers is consistent with the results of Gallagher et al. (13). However, no improvement due to exercise in specific bone sites (spine, radius, and hip) was evident. In fact, exercise resulted in a significant decrease in femoral neck BMD. This observation and geometric changes in the hip due to exercise are discussed elsewhere (8). Examination of regions of interest from DXA suggested that increases in TBBMC at 6 months were associated with gains in the legs. Although Friedlander et al. (11) did not assess TBBMC, exercise training prevented a 3% decrease in spine trabecular BMD in young women. The benefit of exercise on TBBMC is important given that the age range of the subjects studied here (18–31 yr) may be less responsive to interventions aimed at changing bone measures. At this age, modeling has largely ceased to increase bone strength and mass, and the decrease in muscle strength that leads to remodeling adjustments in bone strength and mass after 30 yr of age has not yet begun to a great extent. According to Frost’s (12) biomechanical model, muscle forces of contraction comprise the dominant skeletal loads, which turn bone modeling/remodeling on or off.

Cross-sectional studies show that exercising adults have an advantage in bone mass of 6–20% compared with nonexercising adults (8), but these studies fail to account for selection bias, exercise history, and other confounding variables. The few randomized exercise studies of young women have reported mixed results possibly due to differences in mode of training, compliance, length of study, age of subjects, and hormonal status (3,11,18,29).

The nature of mechanical loading has been postulated to influence the impact of exercise training on bone measures. The studies using only resistance training (18,27) have reported contrasting effects on bone. Resistance exercises have much lower peak stress or strain rates on bones and significant stress magnitudes at harmonic frequencies below 2 Hz, which is in contrast to those of up to 15 Hz in the current study due to the inclusion of jumping rope (1). A 15% increase in loading strain (1900–2000 με) in upper limb bones during a 12-month nonrandomized intervention in 13 women age 19–37 produced flexion strength increases of 14% with no change in bone measures (16). Estimates of peak strains in the present study ranged from 1000 to 6000 με compression (1), which approaches the yield strength for compact bone of 6000–8000 με compression (24). Minimum strains of 1500–2500 με exceed those required for bone maintenance and provide stimulus for bone modeling (35). The Snow-Harter et al. (29) intervention trial employed high-impact running exercise, which would have produced similar strain rates as the current study, but had no greater impact on lumbar BMD than resistance training. Thus, the daily cyclic stress magnitudes and the number of daily loading cycles do not appear to determine bone mass or density changes. In our study, neither changes in fitness, lean body mass, nor self-reported compliance predicted changes in bone measures.

The length of the study relative to the number of remodeling cycles is important because of the lag between bone resorption and bone formation. The short-term studies that have demonstrated gains in bone mass in less than 1 yr may only reflect the remodeling transient (29). The long-term effect is better shown in year 2 and beyond in intervention studies. In the current study, TBBMC appeared to increase beyond the first 6 months, in contrast to the findings of Lohman et al. (18).

Age of subjects does not adequately explain differences among studies. Exercise increased bone density in the approximately 20-yr-olds in the Snow-Harter et al. study (29), in 28- to 39-yr-old women in the Lohman et al. (18) study, in premenopausal women (mean age 37 yr) (3), but not in the 30- to 40-yr-old subjects in the Sinaki et al. (27) study or postmenopausal women (mean age 54) (3). On the other hand, jumping to an average height of 8.5 cm about 2 min a day 6 d·wk-1 significantly increased femoral BMD in women aged 36.4 ± 7.6 yr but not in postmenopausal women even though mean ground reaction forces were 3 times body weight for the premenopausal women and 4 times body weight for the postmenopausal women (3). Younger age groups may be more likely to respond to exercise intervention during more active phases of bone modeling, especially at some skeletal sites. A retrospective analysis of physical activity of the women in the current study showed that high school athletic participation and occupational and leisure energy expenditure for 5 yr before the study were significant predictors of TB and spine BMC or BMD, but only high school athletic participation predicted femoral neck BMD (32). This is consistent with reports that peak bone mass in the femoral neck is attained by age 16 (34). The variable response of exercise on total body and various sites in our study may relate to the age-related variation in achieving peak bone mass. Total body and spine were more responsive than femoral neck and the benefit of exercise on the former was greater in younger women.

Reports of the impact of OCont use on BMD in premenopausal women have been varied (14,15,17,19,20). In the current study, OCont use had a negative impact on TBBMC compared with no OCont use. However, there was no exercise/OCont interaction on TBBMC. The most surprising result of the current study was the negative interaction between exercise and OCont use in spine BMC and BMD, which became apparent during the first 6 months of the intervention. Further, the exercise and OCont group spine BMC and BMD remained significantly lower for the 2 yr of intervention. There were also concurrent decreases in serum calcium levels at 6 months in the exercise and OC use group. Similarly, young adult female macaques on OConts, who would presumably be active, had less bone accretion and lower serum calcium levels by 10 months than the nonuse group (23). A recent cross-sectional study of 20- to 35-yr-old women also suggested a negative impact on bone with a combination of long term oral contraceptive use and exercise history (15). Further support is provided by a recent prospective study demonstrating that long-term (5-yr) low-dose contraceptive use prevented an increase in spine BMD in 19- to 22-yr-old women (20). In the macaque study, reduced levels of serum alkaline phosphatase and serum acid phosphatase suggested lower bone turnover due to OCont use (23). Bone turnover, as calculated from bone formation (predicted from log alkaline phosphatase) plus bone resorption (predicted from ln hydroxyproline/creatinine), using equations previously determined in our laboratory (36) in young women, was lower in OCont users in the current study at baseline. Thus, OCont users began the study at a different state of remodeling that may have led to a differential response to exercise. Bonen et al. (7) observed that in non-OCont users, estradiol (a suppressor of bone remodeling) increases in response to exercise but not in OCont users.

Individuals in a reduced state of bone turnover such as OCont users are not expected to lose bone (10). Only two explanations are plausible for loss of bone. Either an imbalance of resorption and formation occurred or newly formed bone was not as well mineralized. We were not able to detect consistent changes in biochemical markers of bone turnover that accompanied loss of bone in the spine in the exercisers who were OCont users. If mineralization was inadequate, the nutrient most likely deficient in this population is calcium. A calcium by exercise interaction was reported in an analysis of 16 studies in postmenopausal women by Specker (30) such that exercise intervention resulted in increased spine BMD only in women consuming calcium intakes greater than 1 g·d-1. For this reason, calcium supplementation was included in the Lohman et al. (18) and Sinaki et al. (27) studies, with varied results. In the current study, only three women in the exercise and OCont use group had calcium intakes of greater than 1200 mg·d-1. The mean percent change of spine BMC of these three subjects at 6 months was 2.63%. Thus, they were protected from the negative interaction of OCont use and exercise. It seems that adequate mineral must be present for bone modeling to occur under the stimulus of mechanical loading. Yet, calcium supplementation failed to produce a benefit of exercise in the randomized trials of Friedlander et al. (11) and Sinaki et al. (27).

Strengths of this study include the 2-yr randomized exercise intervention, examination of the effects of age range and OCont use, quantifying forces specific to exercises used in our intervention, repeated analysis of diet and continuous analysis of physical activity, and detailed analysis of subject characteristics. Limitations of this study are the poor compliance to exercise and retention. Although using a supervised exercise intervention may improve compliance and retention in the context of a research study, our results better reflect what might occur in response to public health exercise recommendations given to sedentary, young women.

In summary, a 2-yr exercise intervention increased TBBMC, whereas nonexercisers decreased TBBMC. In contrast, exercisers had a decrease in femoral neck BMD, whereas nonexercisers had no change in femoral neck BMD. The unexpected negative impact of exercise in combination with OCont use on spine BMC and BMD suggests that exercising young women who take OCont compromise attainment of spinal peak BMC and BMD. OC use and initiating a fitness program are common practices of young women. A negative interaction may be a partial explanation for the greater fracture rates reported among OC users in two large epidemiological studies of premenopausal women (9,38). Future research should examine mechanisms and countermeasures of the observed interaction. 31

This study was supported by a grant from National Institute of Health (Grant number NIAMS RO1-AR-39560).

Address for correspondence: Connie M. Weaver, Ph.D., Department of Foods and Nutrition, Purdue University, 1264 Stone Hall, West Lafayette, IN 47907-1264; E-mail: [email protected]


1. Anderson, D. D., B. M. Hillberry, D. Teegarden, W. R. Proulx, C. M. Weaver, and T. Yoshikawa. Biomechanical analysis of an exercise program for the forces and stresses in the hip joint and femoral neck. J. Appl. Biomech. 12: 292–312, 1996.
2. Bailey, D. A., R. Faulkner, and H. A. McKay. Growth, physical activity, and bone mineral acquisition. Exec. Sport Sci. Rev. 24: 233–267, 1996.
3. Bassey, E. J., M. C. Rothwell, J. J. Littlewood, and D. W. Pye. Pre- and postmenopausal women have different bone mineral density responses to the same high-impact exercise. J. Bone Miner. Res. 13: 1805–1813, 1998.
4. Bemben, D. A., Boileau, R. A., Bahr, J. M., Nelson, and J. B. Misner. Effects of oral contraceptives on hormonal and metabolic responses during exercise. Med. Sci. Sports Exerc. 24:434–441, 1992.
5. Block, G., M. Woods, H. Potosky, and C. Clifford. Validation of a self-administered diet history questionnaire using multiple diet records. J. Clin. Epidemiol. 43: 1327–1335, 1998.
6. Block, J. N., R. Smith, A. Friedlander, and H. K. Genant. Preventing osteoporosis with exercise: a review with emphasis on methodology. Med. Hypotheses 30: 9–19, 1989.
7. Bonen, A., F. W. Haynes, and T. E. Graham. Substrate and hormonal responses to exercise in women using oral contraceptives. J. Appl. Physiol. 70: 1917–1927, 1991.
8. Burr, D. B., T. Yoshikawa, D. Teegarden, G. P. Mccabe, L. D. Mccabe, and C. M. Weaver. Exercise and oral contraceptive use suppress the normal age-related increase in bone mass and strength of the femoral neck in women 18–31 years old. Bone 27: 855–863, 2000.
9. Cooper, C., P. Hannaford, P. Croft, and C. R. Kay. Oral contraceptive pill use and fractures in women: a prospective study. Bone 14: 41–45, 1993.
10. Forwood, M. R., and D. B. Burr. Physical activity and bone mass: Exercise in futility? Bone Miner. 21: 89–112, 1993.
11. Friedlander, A. L., H. K. Genant S. Sadowsky N. N. Byl, and C-C. Gluer. A two-year program of aerobics and weight training enhances bone mineral density of young women. J. Bone Miner. Res. 10:574–585, 1995.
12. Frost, H. M. On our age-related bone loss: insights from a new paradigm. J. Bone Miner. Res. 12: 1539–1546, 1997.
13. Gallagher, J. C., D. Goldgar, and A. Moy. Total bone calcium in normal women: effect of age and menopause status. J. Bone Miner. Res. 2: 491–496, 1987.
14. Garnero, P., E. Sornay-Rendu, and P. D. Delmas. Decreased bone turnover in oral contraceptive users. Bone 16: 499–503, 1995.
15. Hartard, M., P. Botterman, P. Bartenstein, D. Jeschke, and M. Schwaiger. Effects on bone mineral density of low-dose contraceptives compared to and combined with physical activity. Contraception 55: 87–90, 1997.
16. Heinonen, A., H. Sievanen, P. Kannus, P. Oja, and I. Vuori. Effects of unilateral strength training and detraining on bone mineral mass and estimated mechanical characteristics of the upper limb bones in young women. J. Bone Miner. Res. 11: 490–501, 1996.
17. Kleerekoper, M., R. S. Brienza, L. R. Schultz, and C. C. Johnson. Oral contraceptive use may protect against low bone mass. Arch. Intern. Med. 151: 1971–1976, 1991.
18. Lohman, R., S. Going, R. Pamenter, et al. Effects of resistance training on regional and total bone mineral density in premenopausal women: a randomized prospective study. J. Bone Miner. Res. 10: 1015–1024, 1995.
19. Mazess, R. P., 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.
20. Polatti, F., F. Perotti, N. Filippa, D. Gallina, and R. E. Nappi. Bone mass and long-term monophasic oral contraceptive treatment in young women. Contraception 51: 221–224, 1995.
21. Raisz, L. G., B. Writa, A. Artis, et al. Comparison of the effects of estrogen alone plus androgen on biochemical markers of bone formation and resorption in postmenopausal women. J. Clin. Endocrinol. Metab. 81: 37–43, 1996.
22. Recker, R. R., K. M. Davies, S. M. Hinders, R. P. Heaney, R. R. Stegman, and D. R. Kimmel. Bone gain in young adult women. JAMA 268: 2403–2408, 1992.
23. Register, T. C., M. J. Jayo, and C. P. Jerome. Oral contraceptive treatment inhibits the normal acquisition of bone mineral in skeletally immature young adult female monkeys. Osteoporos. Int. 7: 348–353, 1997.
24. Reilly, D. T., and A. H. Burstein. The elastic and ultimate properties of compact bone tissue. J. Biomech. 8: 393–405, 1975.
25. Rockwell, J. C., A. M. Sorensen, S. Baker, et al. Weight training decreases vertebral bone density in premenopausal women: a prospective study. J. Clin. Endocrinol. Metab. 71: 988–993, 1990.
26. SAS Institute Inc . SAS/Stat User’s Guide, Version 6, 4th Ed., Vols. 1 and 2, Cary, NC: SAS Institute, 1989.
27. Sinaki, M., H. W. Wahner, E. J. Bergstrahl, et al. Three-year controlled, randomized trial of the effect of dose-specified loading and strengthening exercises on bone mineral density of spine and femur in nonathletic, physically active women. Bone 19: 233–244, 1996.
28. Slemenda, C. W., T. K. Reister, S. L. Hui, J. Z. Miller, J. C. Christian, and C. C. Johnston, Jr. Influences on skeletal mineralization in children and adolescents: evidence for varying effects of sexual maturation and physical activity. J. Pediatr. 125: 201–207, 1994.
29. 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.
30. Specker, B. L. Evidence for an interaction between calcium intake and physical activity on changes in bone mineral density. J. Bone Miner. Res. 11: 1539–1544, 1996.
31. Specker, B. L., L. Mulligan, and M. Ho. Longitudinal study of calcium intake, physical activity, and bone mineral content in infants 6–18 months of age. J. Bone Miner. Res. 14: 569–576, 1999.
32. Teegarden, D., W. R. Proulx, M. Kern, et al. Previous activity relates to bone mineral measures in young women. Med. Sci. Sports Exerc. 28: 105–113, 1996.
33. Teegarden, D., W. R. Proulx, B. R. Martin, et al. Peak bone mass in young women. J. Bone Miner. Res. 10: 711–715, 1996.
34. Theintz, G., B. Buchs, R. Rizzoli, et al. Longitudinal monitoring of bone mass accumulation in healthy adolescents: evidence for marked reduction after 16 years of age at the levels of lumbar spine and femoral neck in female subjects. J. Clin. Endocrinol. Metabol. 75: 1060–1065, 1992.
35. Turner, C. H. Homeostatic control of bone structure: an application of feedback theory. Bone 12: 203–217, 1991.
36. Weaver, C. M., M. Peacock, B. R. Martin, et al. Quantification of biochemical markers of bone turnover by kinetic measures of bone formation and resorption in young healthy females. J. Bone Miner. Res. 12: 1714–1720, 1997.
37. Welton, D. C., H. C. G. Kemper, G. B. Post, and W. A. Vanstaveren. A meta-analysis of the effect of calcium intake on bone mass in young and middle aged females and males. J. Nutr. 125: 2802–2813, 1995.
38. Vessey, C., J. Mant, and R. Painter. Oral contraceptive and other factors in relation to hospital referral for fracture-findings in a large cohort study. Contraception 57: 231–235, 1998.


© 2001 Lippincott Williams & Wilkins, Inc.