Effect of Two Training Regimens on Bone Mineral Density in Healthy Perimenopausal Women: A Randomized Controlled Trial



The aim of this randomized controlled trial was to evaluate the effects of 18 months of calisthenics and endurance training regimens on bone mineral density (BMD) in perimenopausal women. Clinically healthy sedentary female volunteers (n = 105) aged 52–53 years were randomly assigned to a calisthenics (n = 36), endurance (n = 34), or control (n = 35) group. The calisthenics training (2.6 times per week on average, 50 minutes per session) consisted of rhythmic strength-endurance exercises by large muscle groups, and the endurance training (3.2 times per week, 50 minutes) consisted of walking, stair climbing, ergometer cycling, and jogging at a controlled heart rate zone corresponding to 55–75% of the individual maximal oxygen uptake (VO2max) of the subjects. The control subjects performed a light stretching program once a week. The BMD of the lumbar spine (L2-L4), right femoral neck, calcaneus, and distal radius was measured by dual-energy X-ray absorptiometry at 0, 4, 8, 10, 14, and 18 months, and the maximal isometric strength during trunk extension and flexion, leg extension, and arm flexion and the VO2max by ergospirometry were evaluated at 0, 8, 10, and 18 months of intervention. The VO2max improved significantly (p = 0.021) in the endurance group. The linear trend of the femoral neck BMD in the endurance group, as determined by generalized linear models, was significantly different (p = 0.043) from that of the control group, the trend indicating a maintenance of the prestudy BMD. In the calisthenics group, the training effect was not significant. However, the distal radius BMD of the endurance group showed a significant negative trend (p = 0.006). These results suggest that multiexercise endurance training maintains the BMD the clinically important femoral neck of perimenopausal women. This form of endurance training proved also to be feasible for healthy perimenopausal women.


Hip fractures among the elderly are a significant public health problem, and their number is expected to rise dramatically as populations age.(1) The prevailing view is that one of the major determinants of fracture risk is bone mineral density (BMD).(2) Several studies have reported a high BMD in physically active subjects as compared with sedentary subjects in both cross-sectional(3,4) and prospective studies on premenopausal(5–7) and postmenopausal women.(8–10) Thus, physical activity seems to be effective in maintaining and even increasing bone mass and strength.(11)

It is known that the skeletal response to loading is characteristic of different age phases, and that the skeleton of growing girls and premenopausal, perimenopausal, and postmenopausal women can respond differently(12) and in a site-specific fashion.(13–15) A few studies have assessed the BMD response to exercise at the clinically important femoral neck in postmenopausal women.(8–10) Corresponding information concerning perimenopausal women about to enter menopause is lacking. The skeleton of women at that stage may be more inclined to maintain bone mass with physical activity than later in the presence of postmenopausal estrogen deficiency.

The purpose of this study was to evaluate the effects of calisthenics and endurance training regimens on BMD at several skeletal sites in perimenopausal women. We recently reported significant BMD gains in an 18-month, high-impact exercise regimen in premenopausal women.(5) However, high-impact exercise might be too rigorous or even injurious to start at an older age especially for women with a sedentary background. Consequently, the rationale of using the above training regimen was based on the findings suggesting that even a relatively light weight–bearing exercise may to be beneficial and feasible in elderly sedentary women who have a low activity level, and, in addition, endurance training and resistance training have been shown to be effective in all adult groups.(7–12,16–22) Therefore, endurance training and calisthenics were assumed to be a feasible and effective means to promote the health of bones in elderly individuals.



The study was a randomized controlled trial with three experimental groups: an endurance training group, a calisthenics training group, and a control group. The intended training frequency was four times a week for a period of 18 months. Measurements were done at baseline and at 8, 10, and 18 months, except for the BMD measurements, which were made also at 4 and 14 months in order to follow the time course of changes in BMD in a more specific fashion.


A questionnaire was sent to a random population sample of 1242 women from a cohort of 51- to 53-year-old women (n = 3146) living in the city of Tampere asking if they were interested in participating in the study. Five hundred and sixty-nine women expressed their interest. One hundred and forty-one potentially eligible women were invited to the screening examination (Fig. 1). The criteria for the invitation were being willing to participate, clinically healthy (no cardiovascular, musculoskeletal, respiratory, or other chronic diseases that might limit training or testing), sedentary (vigorous exercise no more than twice a week), not excessively obese (body mass index <33), nondieting, nonsmoking, and no apparent occupational or leisure time responsibilities that might impede participation. At the medical screening examination, 104 women had normal resting and exercise electrocardiogram (ECG) and were selected for the study. At this point, 3 women were no longer willing to participate, and thus 101 women started the study. Random allocation assigned 32 women to the endurance training group, 35 women to the calisthenics group, and 34 women to the control group. Baseline characteristics of the subjects are given in Table 1.

Figure FIG. 1.

Trial profile.

Table Table 1. Baseline Characteristics and Calcium Intake of the Subjects
original image

Estrogen status

The estrogen status was obtained by interviewing each subject on her use of estrogen supplements and concurrent menstrual status. The women were classified into two groups: the nonoestrogen category consisting of subjects who were post- or perimenopausal (menstrual irregularities or menstrual function cessation during the 6 months before the study) and did not use estrogen replacement; and the estrogen category, consisting of women who had normal menstruation or used estrogen replacement (Table 1). The estrogen status of some women changed during the study as follows: one woman each in the endurance and calisthenics groups and two women in the control group became postmenopausal (menstrual function ceased) and one woman in each group started estrogen replacement therapy.

Calcium intake

Dietary food intake, including calcium, and the use of vitamin and mineral supplements were estimated from a 7-day food record at baseline, and after 8 and 18 months.

Bone mineral density

Areal BMD (g/cm2) was measured at lumbar spine (L2-L4), the right femoral neck, the calcaneus, and the dominant distal radius with dual-energy X-ray absorptiometry (Norland XR–26, Norland Corp., Fort Atkinson, WI, U.S.A.). All the scanning and analyses were done by the same operator according to our established procedures.(23) The in vivo day-to-day precision (coefficient of variation) of the BMD measurement in our laboratory ranges from 0.7 to 1.7%. The scanner was calibrated daily, and its performance was followed with our quality assurance protocol.(24) There was no significant machine drift during the study period.

Cardiorespiratory performance

After a standard resting ECG (Marquette Case 12; Marquette Electronics, Milwaukee, WI, U.S.A.), an incremental bicycle ergometer test (Siemens Elema RE 820, Rodby Elektronik AB, Enhörna, Sweden) with step increments of 10 W/minute was conducted to volitional maximum. Maximal oxygen consumption (VO2max) was measured with an automatic metabolic analyzer (Sensor Medics 2900Z; SensorMedics, Anaheim, CA, U.S.A.). The following criteria for the maximality of the test were used: unable to continue to next load, heart rate (HR) >85% of age-predicted maximum, respiratory quotient >1.1, and no marked increase in VO2max during the last minute. The exercise ECG was monitored throughout the test (Marquette Case 12; Marquette Electronics), the systolic blood pressure was measured, and Borg's rating of perceived exertion (scale 6–20) was assessed at 2-minute intervals under a physician's supervision.

Muscle strength

The maximum isometric strength of the trunk extensors and flexors and dominant elbow flexors was measured with an isometric arm dynamometer (Digitest, Muurame, Finland), and the strength of the leg extensors with an isometric leg press dynamometer (Tamtron, Tampere, Finland). The measurement procedures have been described in detail elsewhere.(25) The dynamometers were calibrated on a monthly basis; there was no drift.

Exercise programs

The subjects in the endurance and calisthenics groups were asked to exercise four times per week for 18 months. During the first month, the subjects accustomed themselves to the training. The bout duration was 50 minutes (10 minutes warm-up, 30 minutes effective, 10 minutes cool-down). At least once a week the exercise session was supervised by a professional exercise leader.

The endurance training included intervals of walking, jogging, ergometry cycling, stair-climbing, and graded treadmill exercises at 55–75% of the VO2max. The percentage of VO2max was individually assessed on the basis of the ergospirometric VO2/HR data and the exercise HR zone were controlled by a portable HR meter (Polar Edge; Polar Electro, Kempele, Finland).

The calisthenics training consisted of a warm-up period including stretching and balance exercises using dance steps with music. The effective period consisted of 3 sets of 16 repetitions of 8 different rhythmic muscular strength-endurance calisthenics exercises designed to load the muscles of the trunk, pelvis, hip, and lower limbs with ordinary pace (hip flexion-extension, abduction-adduction, knee extension-flexion, sit-ups, trunk extension and rotations, and upper limb movements). The subjects used additional wrist and ankle weight bands (1–2 kg).

The control group did stretching exercises once a week and were otherwise asked to maintain their prestudy level of physical activity. This sham exercise was designed to standardize the exercise habits and to maintain the control group's interest in the study.

The actual training intensity was monitored with continuous HR recordings (Sport Tester 3000; Polar Electro) during the endurance and calisthenics training sessions of 1 training week in every 4 months of the program. In addition, all the subjects kept a continual exercise diary with consecutive 7-day forms for type, duration, and self-assessed intensity of all physical activities. The diaries were returned and checked for completeness once a month.

Statistical methods

General linear models (GLMs) with restricted maximum likelihood estimation were used to assess the effects of exercise intervention on BMD, muscular strength, and cardiorespiratory fitness.(26) The GLM analysis allows incorporation of incomplete data into the models. For BMD, the training effect was determined as the difference between the training and control groups, the difference being evaluated in two ways: the post-training difference and the difference in the slopes of linear trend.

The rationale of presenting the linear trend analysis in addition to the conventional post-training difference was that we had BMD data from six measurement points. These longitudinal BMD data provided the possibility of taking the time course of the BMD response into account. The linear trend analysis is, by nature, less sensitive to random variability and was thus anticipated to give a more reliable estimate of the actual trend than the end-point analysis. For the muscle strength and cardiorespiratory fitness, only the post-training difference was determined. The possible confounding variables considered in this study were weight, calcium intake, estrogen status, and exercise adherence. Of these variables, only the estrogen status had a confounding effect on BMD, and, therefore, the BMD differences were adjusted for estrogen status and baseline values. Differences in cardiorespiratory fitness and muscle strength, however, were adjusted only for baseline.


Program adherence and compliance

A total of 27 subjects in the control group, 26 in the calisthenics group, and 23 in the endurance group completed the study. Twenty-five (25%) of the subjects dropped out for the following reasons: overuse injury (1 in the control group and 1 in the endurance group), death from cancer (1 in the control group), change of residence (1 in the calisthenics group), and loss of interest or increased occupational responsibilities or insufficient time (21 subjects). The drop-out rate was equal in the experimental groups.

The reported exercise program compliance, defined as the percentage of the completed weekly exercise sessions out of the prescribed sessions, averaged 72% (0.72 times) for the control group, 66% (2.6 times) for the calisthenics group, and 80% (3.2 times) in the endurance group. The mean exercise HR was 108 (SD 8) beats/minute in the calisthenics group and 141 (SD 8) beats/minute in the endurance group. The exercise intensity in terms of maximal VO2max was estimated (on the basis of measured HR) as 43% (SD 11) in the calisthenics group and 72% (SD 7) in the endurance group. The proportion of sessions attended did not correlate with the BMD changes at any bone site.

Calcium intake

Table 1 includes the baseline levels of daily calcium intake. The calcium intake appeared sufficient in all groups, especially in light of the fact that the 7-day food record has been shown to underestimate the true calcium intake.(27) There was no change in the calcium intake among the groups during the study period.

Changes in BMD

The baseline BMD values are given in the Table 1 and the training effects in Table 2. For the BMD of the femoral neck, the linear trend in the endurance group was significantly different (p = 0.043 in the GLM analysis) from that of the control group, with the trend indicating a maintenance of the prestudy BMD level. In the calisthenics group, the training effect was not significant. For the BMD of the lumbar spine, calcaneus, and distal radius, there were no significant training effects in either the endurance or the calisthenics group, except for the distal radius, for which the linear trend (p = 0.006 in the GLM analysis) and post-training difference (p = 0.007 in the GLM analysis) were significantly lower in the endurance group than in the control group. These sites showed a clear negative trend as compared with corresponding control data. The time courses of the change in the BMD are presented in Fig. 2 for the training groups and the control group.

Figure FIG. 2.

The estimated changes in the BMD (GLM analysis) at (A) the lumbar spine, (B) the femoral neck, (C) the calcaneus, and (D) the distal radius during the study period. The solid line represents the endurance group, the dashed line the calisthenics group, and the dotted line the control group. Bars indicate standard error of estimates.

Table Table 2. Training Effects on BMD Was Determined as a Training Group Difference from the Control Group
original image

Cardiorespiratory fitness and maximal isometric strength

The training effect expressed as post-training changes of the cardiorespiratory fitness and muscular strength are shown in Table 3. Among women who completed the study, the VO2max improved 15% in the endurance group, 11% in the calisthenics group, and 9% in the control group. The post-training mean VO2max was significantly higher (p = 0.021 in the GLM analysis) in the endurance group than that in the control group.

Table Table 3. Training Effects on Muscular Strength and Cardiorespiratory Fitness Was Determined as a Training Group Difference from the Control Group
original image

There were no significant differences between the trainees and the controls in the change of muscular strength (Table 3).


In this study, the multiexercise endurance training at a relatively high intensity (about 70% of VO2max) maintained BMD of the clinically important femoral neck of healthy perimenopausal women. However, no training effects in BMD occurred at other sites in the endurance training group and at no site in the calisthenics group. In the control group, BMD decreased at all sites in the anticipated age-related manner.

The osteogenic stimulus controlling bone mass originates from the loading-induced mechanical deformation (strain). It has been suggested that skeletal remodeling is related to “error signal” (the difference between local strain caused by habitual and specific loading), and consequently the response is site specific.(28) Our previous data suggest that training producing strain at a high rate and with high peak forces in versatile movements is effective osteogenic stimulus in young women.(3,5,29) Even fast walking may produce effective osteogenic stimulus for the lower limb skeleton by altering the loading direction and rate by a shorter heel strike and higher ground and hip reaction forces.(30–32) Anderson et al.(33) found that the magnitudes of peak stress or strain rate were significantly higher for walking, jumping rope, and jogging than for weight-lifting exercises.(33) In fact, strain in bone tissue has been shown to be proportional to the speed of walking.(34) Consequently, fast walking, uphill walking, and slow jogging, as used in the endurance program of this study, were likely to produce higher strain rates and different strain distributions as compared with customary walking during normal daily routines. In contrast, the BMD changes in the calisthenics group did not differ from the values of the control group, probably because of slow and too pliant training movements. In other words, the present calisthenics program apparently did not produce a sufficiently high strain rate and uncustomary strain distribution to result in an effective loading stimulus. It is noted that any exercise regimen for influence on skeletal mass must induce a change in loading at all skeletal sites of interest.

An apparent explanation for the positive effect on the BMD of the femoral neck in the endurance group was the nature of the training-induced loading. Brisk and intense walking and jogging, which were the main components of the endurance training, can produce ground reaction forces 1.2–1.5 times body weight when walking at ordinary pace(31) and 2–5 times body weight when running.(32) Due to bad lever arms of muscles involved,(35) the forces acting across the hip joint during even the ordinary walking can be up to 4.8 times the body weight.(30) Thus, during fast walking or jogging, the lower limbs are expected to experience loading, whose magnitude and rate can be quite high, and the effect may be enhanced by repetitive cycles. In this light, it was somewhat surprising that there was no effect on the BMD of the calcaneus. Probably the loading pattern of this peripheral site was not altered by intense walking as much as was the loading environment of the femoral neck.

The mean exercise intensity in the endurance group was about 70% of the VO2max, a level probably approaching the anaerobic threshold of the subjects. If so, the walking speed was probably near to the limit when the subject would shift from walking to jogging. This “borderline” situation of movement may have modified further the bone loading environment as compared with more customary movements (e.g., slow walking or jogging). This explanation is in line with that of Hatori et al.(16) who showed in postmenopausal women that lumbar BMD increased with very fast walking (7.2 km/h) performed at intensities exceeding the anaerobic threshold. In contrast, slower walking (6.2 km/h) at intensities below the threshold was not sufficient to increase lumbar BMD. Likewise, Martin and Notelovitz(17) observed that walking speeds of less than 6.4 km/h did not increase lumbar BMD. This was further confirmed by the findings of Dalsky et al.(18) and Chow et al.(19) who suggested that high intensity exercise (70–90% of VO2max) can increase or maintain BMD in postmenopausal women. Unfortunately, femoral neck BMD was not measured in these studies, and thus it remains unclear whether there was a training effect on this clinically relevant site.

Although the endurance training slowed the age-related femoral bone loss in this study, it appeared to increase slightly the radial bone loss. We have observed a similar trend in premenopausal women's radial BMD in our recent 18-month high-impact exercise study.(5) Likewise, athletes whose training did not intensively load the upper limbs seemed to have a lower radial BMD as compared with sedentary referents.(3) These observations are suggestive of a “steal phenomenon,” or redistribution of bone mineral from nonloaded sites to loaded sites. The relative decrease in radial BMD was, however, small (about 2%) and thus would not account for a substantial increase in the risk for a forearm fracture. Moreover, forearm fractures are rarely fatal, and they cause much less disability and expense than hip fractures.(36) However, as regards hip fracture, the age of the present study group (about 20 years before the age of increased risk)(36) and the short follow-up time of the subjects in conjunction with the relatively small, though significant, training effect on femoral neck did not allow us to extrapolate the actual role of endurance training in preventing these fractures in later life.

In general, exercise has been shown to have a positive effect on BMD at various skeletal sites in both premenopausal(5–7,20,21,37) and postmenopausal women.(8–10,12,16–19) Bassey and Ramsdale(38) suggested, in their recent study, that the sensitivity of the skeleton to exercise intervention may even increase rather than decrease as the time since menopause increases. Lane et al.(39) reported a remarkable difference in lumbar BMD between an exercise and a control group with a history of about 10 years of running, which was started just before or during early menopause (i.e., at the perimenopausal phase). They suggested that the bone maintenance effect of exercise during the perimenopausal period may be an essential factor, making a favorable difference as compared with sedentary women later in life.

As regards the general acceptability of the present endurance training regimen, walking and jogging are natural ways of human movement, safe, inexpensive, and easy to start and train. These facts imply a good adoption of this kind of training among healthy subjects as a part their everyday life. Moreover, not only the shown skeletal benefits but also many other apparent health benefits(40) speak in favor of the general effectiveness of endurance training as a public health measure. However, the relatively high training intensity (about 70% of the VO2max) required for the training effect, the continuous nature of the training, and the total time needed (e.g., for weekly exercise sessions) together may lead to reduced long-term feasibility of the regimen.

In conclusion, the present study showed that an 18-month program of relatively intense, multiexercise endurance training resulted in maintenance of BMD at the clinically important femoral neck of perimenopausal women. This form of endurance training proved also to be safe and feasible among this subject group. Further longitudinal studies are needed to determine whether endurance training would offer an effective means to prevent femoral neck fractures in later life.


The authors thank Virpi Koskue for her expert dual-energy X-ray absorptiometry measurements. Financial support was obtained from The Ministry of Education in Finland.