JOURNAL OF BONE AND MINERAL RESEARCH Volume 7, Number 7, 1992 Mary Ann Liebert, Inc., Publishers
Effects of Resistance and Endurance Exercise on Bone Mineral Status of Young Women: A Randomized Exercise Intervention Trial CHRISTINE SNOW-HARTER,* MARY L. BOUXSEIN, BARBARA T. LEWIS, DENNIS R. CARTER, and ROBERT MARCUS
ABSTRACT A substantial body of cross-sectional data and a smaller number of intervention trials generally justify optimism that regular physical activity benefits the skeleton. We conducted an 8 month controlled exercise trial in a group of healthy college women (mean age = 19.9 years) who were randomly assigned to a control group or to progressive training in jogging or weight lifting. We measured the following variables: bone mineral density (BMD) of the spine (L2-4) and right proximal femur using dual-energy x-ray absorptiometry, dynamic muscle strength using the 1-RM method, and endurance performance using the 1.5 mile walk/run field test. A total of 31 women completed the 8 month study. For women completing the study, compliance, defined as the percentage of workout sessions attended, was 97% for the runners (range 90-100%) and 92% (range 88-lOoV0) for the weight trainers. Body weight increased by approximately 2 kg in all groups 0, < 0.05). Weight training was associated with significant increases (p < 0.01) in muscle strength in all muscle groups. Improvement ranged from 10% for the deep back to 54% for the leg. No significant changes in strength scores were observed in the control or running groups. Aerobic performance improved only in the running group (l6V0, p < 0.01). Lumbar BMD increased (p < 0.05) in both runners (1.3 f 1.6%) and weight trainers (1.2 f 1.8%). These results did not differ from each other but were both significantiy greater than results in control subjects, in whom bone mineral did not change. N o measure of bone mineral at the proximal femur changed significantly in any group. These results demonstrate that 8 months of supervised progressive training in either running or resistance exercise modestly increases lumbar spine mineral in young women.
INTRODUCTION justifies optimism for the concept that regular physical exercise benefits the skeleton. Exercise prescriptions may be defined according to type, intensity, duration, and frequency, but we do not yet understand the details of a program that would be optimal for bone.'') Bone adapts to the loads applied to it. This adaptation includes an increase in density when mechanical loading increases and a loss of bone density when customary loads are removed. This concept has been expanded
C
ONSIDERABLE EVIDENCE
to a general theory of bone remodeling in which mechanical loading is a primary regulator. ('-') The mechanisms by which mechanical forces alter bone mass await clarification, but Carter et al.(2)proposed that the apparent density of a bone region is determined by its customary loading history. To expand this concept, Whalen et al.'s' proposed a mathematical model that relates bone mass at equilibrium to its typical loading experience. This daily history is defined as the sum of a series of loading events, each characterized by a load magnitude and number of cycles. The studies of Whalen et aI.(') suggest that the ef-
Musculoskeletal Research Laboratory, GRECC, Department of Veterans Affairs Medical Center, Palo Alto, and the Departments of Medicine and Mechanical Engineering, Stanford University, Stanford, California. *Current address: Department of Exercise and Sports Science, Oregon State University, Corvallis.
761
SNOW-HARTER ET AL.
762 fect of load magnitude should outweigh that of cycle number. If this formulation is correct, it would be a reasonable hypothesis that muscle strength training, an activity of high loads and few repetitions, would have a greater effect on bone mass than endurance exercise, a high repetitionlow magnitude activity. In this regard, Heinrich et a1.‘61 found that axial and peripheral bone mass were higher in young women who participated in weight lifting exercise than in endurance-trained women of similar age. We conducted a controlled exercise intervention trial in a group of healthy college women who were randomly assigned to a control group or to a program of progressive training in jogging or weight lifting. This study addressed the following two hypotheses: (1) exercise training increases bone mineral content of the lumbar spine and proximal femur, and (2) changes in spine and proximal femur mineral content are greater with weight training than with jogging. In this paper we present the experimental results of this study.
MATERIALS AND METHODS Subjects A total of 52 women, age 19.9 f 0.7 years (mean f standard error of the mean, SEM), were recruited from the student body of Stanford University. Volunteers who were in good health with regular menstrual cycles (8-12 periods per year) for at least 3 years before the study and who were without chronic physical problems that would limit participation in an exercise program were invited to participate in this study. Competitive athletes were not invited to enroll, and no subject belonged to an athletic team. On acceptance into the study, subjects were randomly assigned to control, weight training, or running groups. All baseline measurements were completed before initiating the exercise program. Subjects participated in the exercise sessions, completed activity records and exercise logs, and attended group meetings. We defined satisfactory participation in the exercise sessions as attending at least 85% of all sessions and completing >90% of the designated workout at each session attended. All testing and training sessions were conducted under supervision of the research staff. The protocol was approved by the Human Subjects Committee, Stanford University, and all subjects gave written consent.
Bone mineral evaluation Bone mineral content (BMC, g) and areal bone mineral density (BMD, g/cm’) of the lumbar spine (L2-4) and right proximal femur were estimated by dual-energy x-ray absorptiometry (Hologic QDR 1O00, Waltham, MA) on entry and following 8 months of training. The femoral neck, trochanter, and Ward‘s triangle regions of the hip were analyzed separately. In our laboratory, the following precision errors for BMD (coefficients of variation for multiple determinations) have been determined for a group of 20 healthy young adult women: (1) lumbar spine = 0.5%, (2) femoral neck and Ward’s triangle = 0.9%, and (3) trochanter = 0.75%. Bone phantoms (Hologic Corp., Waltham, MA) were scanned each day throughout the study and gave readings that were consistently within 0.7% of nominal values. N o evidence for machine drift appeared during this study.
Muscle strength assessment After formal instruction in the use of weight machines, all subjects underwent an initial assessment of muscle strength. A total of 13 exercises were selected to assess the strength of the arms, chest, shoulder, back, hips, and legs (Table 1). Biceps curl, triceps extension, bench press, military press, upright row, and knee extension and flexion were performed on a Universal Multi-station weight machine (Universal Gym Equipment, Inc., Cedar Rapids, IA). Hip adduction, abduction, extension, and flexion were executed on a Universal Total Hip Trainer model 993494 (Universal Gym Equipment, Inc., Cedar Rapids, IA). Back extension was tested o n the Nautilus Low Back Machine (Nautilus Sports/Medicine Industries, Dallas, TX). Dynamic strength of all major muscle groups was assessed using the one repetition maximum (1 RM) method, as we have validated and reported.‘’) One RM is the maximum weight a subject can lift one time with acceptable form. “Acceptable form” means that performance of the exercise isolated primarily the muscle groups being tested with minimal recruitment of other major muscle groups. When the criterion for acceptable form was not satisfied, the last recorded successful weight was taken for the strength score. The details of each exercise have been de-
TABLE1. EXERCISES USED FOR STRENGTH TESTSAND
Exercise Biceps curl Triceps press Upright row Bench press Sit-ups Military press Lateral pull down
BY
WEIGHT TRAINERS IN WORKOUTSa
Primary muscles
Exercise
Primary muscles
Biceps brachii, brachialis Triceps brachii Deltoids, trapezius Pectoralis major, triceps Rectus abdominus Deltoids, triceps Latissimus dorsi
Knee flexion Back extension Knee extension Hip adduction Hip abduction Hip extension Hip flexion
Hamstrings group Erector spinae Quadriceps group Adductor magnus Gluteus medius Gluteus maximus lliopsoas
alncluded are primary muscles activated to perform exercise.
763
EXERCISE TRAINING IN YOUNG WOMEN scribed previously. (’) Before each strength-testing session, subjects received detailed instructions and performed each exercise two times at very low resistance to enhance familiarization and warm-up. Each 1-RM test began at a weight near the anticipated maximum to minimize fatigue resulting from repetition. In most cases subjects required no more than three repetitions to reach maximum. All tests were repeated with weight increases of 2.5, 5 , and 10 pounds, depending upon the exercise, until the subject was not able to lift additional weight with acceptable form. A 20 s rest period was given between repetitions, with at least 1 minute of rest between exercises. Strength tests were administered on 2 nonconsecutive days, with hip and back strength recorded on 1 day and arm, shoulder, chest, superficial back, and leg strength assessed on the second test date. In our laboratory, the coefficient of variation for repeated measures is 2-6% for all strength tests in young women. These measurements served as baseline values for subjects in the control and running groups as well as for the weight trainers. Muscle strength for the arms, chest, shoulders, hips, back, and legs was recorded as summed values obtained from the 1-RM tests as follows: arms = biceps + triceps press; chest = bench press; shoulders = military press + upright row; superficial back = lateral pull down; deep back = back extension; legs = leg extension + leg curl; hips = average hip abduction + average hip flexion. The term “average” refers to the mean score of the right and left sides. Hip extension and adduction are not included in the expression of hip strength because many subjects had not reached maximum with the entire stack of weight plates.
Cardiorespiratory fitness testing The 1.5 mile walklrun field test was conducted to assess endurance performance.“’ In this simple test, subjects are instructed to travel 1.5 miles on foot as quickly as possible. Any combination of running and walking is permitted. All subjects performed this test at the Stanford University track at the same time of day after a standardized warmup period. Results are scored as the time required to complete the 1.5 mile task. Performance on this test correlates strongly with measures of aerobic fitness obtained from Vozrnax In a pilot study involving 10 young women we found the precision error of this test to be 7%.
Activity assessments O n entry, subjects completed a questionnaire detailing their exercise history during the previous year. Subjects were asked to itemize all recreational activities in which they had participated. They listed the number of hours per week and the number of months per year that they participated in each listed activity. From this questionnaire, the average number of hours of exercise per week was calculated and the women were grouped into three categories: low 5 1 h/week, moderate = 2-3 h/week, and high > 3 h/week. Average daily activity profiles were developed for each
participant from daily activity logs that were collected during the study (October, January, and May). Subjects recorded their activities for 7 consecutive days, registering for each day the number of hours that were spent lying, sitting, standing, walking, and participating in both weight-bearing and non-weight-bearing recreational activities other than those directly related to the training sessions. Mean daily values were calculated for each activity. Subjects received instruction on record maintenance. On submission, records that did not add up to 23-25 h of activity were discarded and subjects were asked to complete records for additional days.
Exercise training protocols This study was designed to correspond to a 35 week school year. Exercise training commenced in early October and finished during the last week in May. Muscle strength, bone mineral, and aerobic fitness assessments were made within 1 week before beginning and following the end of training. Training was interrupted from weeks 12 to 14 during the winter holiday and week 26 during the spring vacation. Although some runners were able to continue exercise, few weight trainers had access to the gymnasium at these times. For women in the weight training group, no reduction in training load occurred following these breaks in training. Weight Training: Subjects performed a circuit of 14 exercises three times per week under supervision. The exercises were identical to those enumerated in Table 1 and have been previously described in detail.(’) At each workout session, the subjects performed all exercises for three sets of 8-12 repetitions each. Training was progressive, with an increase in load of not more than 10% per week. Training began at 6 5 7 0 % of baseline 1 RM values and reached target levels of 75% of the initial 1 RM within 3 months. Strength testing was conducted every 6 weeks throughout the 8 month training period, and training levels were adjusted accordingly. During the last 2 months of the study, training was conducted at 85% of the current 1 RM values. The back extension exercise was the sole exception to this training schedule, since a Nautilus back machine was not available at the training facility. Therefore, to load the deep back musculature, back extensions were performed to exhaustion using lower weights.
Running: Before training was initiated all subjects attended a session in which the following topics were discussed: running form and biomechanics, foot placement and stride, injury prevention, and shoe selection. Subjects ran three times or more each week. At least two-thirds of running sessions were carried out under supervision. Each session began with a warm-up period that included 3-5 minutes of slow jogging followed by stretching and finished with several minutes of walking and stretching. Subjects were instructed to run at a speed sufficient to elevate the heart rate to 70-80% of age-predicted maxima. Running schedules were designed for each subject based on her 1.5 mile walk/run time, and mileage increases were re-
SNOW-HARTER ET AL.
764 stricted to 10% per week. The progression of average weekly mileage was as follows (mean + standard deviation, SD): months 1 and 2 = 4.2 + 2.3 miles, months 3 and 4 = 5.6 + 4.3 miles, months 5 and 6 = 1.9 + 5.1 miles, and months 7 and 8 = 10.5 f 4.9 miles.
Control Group: The control subjects did not change physical activity patterns from baseline. Maintenance of recreational activity at baseline levels was encouraged.
Miscellaneous Meetings were held quarterly for all subjects. Exercise logs were maintained for each workout. For runners these logs included the distance run and the pace. For weight trainers, the number of repetitions and weight lifted were recorded for all exercises. All subjects were issued a daily 500 mg supplement of calcium as carbonate (Calel-D, Rorer Pharmaceuticals, Inc., Fort Washington, PA) for the duration of the study. Incentives were offered to exercising subjects throughout the study period. Awards were given to subjects achieving 80% of stated goals, with additional prizes for exceptional compliance.
Data management and analysis Data were analyzed using the Statview I1 software package (Abacus Concepts, Inc., Berkeley, CA). Analyses included standard descriptive statistics, paired t-tests, and analysis of variance (ANOVA). Changes over time in muscle strength were analyzed as arithmetic before-after differences as well as by percentage differences. Changes in BMD were analyzed by 3 x 2 ANOVA with repeated measures on the trial factor. T h e p values reported for paired ttests were two-tailed. Since bone mineral at the spine was the primary study end point, an a value of 0.05 was taken as the criterion for significance. For other bone sites and for muscle strength changes, (Y = 0.01 was chosen to adjust for multiple comparisons.
RESULTS Subject characteristics (means + SD) on entry are presented in Table 2. No women were obese, as BMI (weight +- height’) was below 30 in all cases. All baseline variables were normally distributed and did not differ significantly among groups, indicating that randomization was effective. A total of 31 women completed the 8 month study, representing an attrition of 40%. Subjects who left the study did so for academic and personal reasons. No subject stopped participation because of injury. One runner developed sustained amenorrhea during the study period. Although she completed the study, we excluded her data from analysis. For women completing the study, compliance, defined as the percentage of workout sessions attended, was 97% for the runners (range 90-100%) and 92% (range 88100%) for the weight trainers. The training programs were well tolerated by all participants. No subject developed
overuse or other injury of sufficient magnitude to require modification or interruption of the training schedule.
Body weight, muscle strength, and endurance performance Pre- and posttraining values are presented in Table 2. Body weight increased by approximately 2 kg in all groups (p < 0.05). Weight training was associated with significant increases (p < 0.01) in muscle strength in all muscle groups (Table 2 and Fig. 1). The magnitude of improvement ranged from 10% for the deep back muscles to 54% for the leg (knee flexors and extensors). No significant changes in strength scores were observed in the control or running groups. Endurance performance, as measured by the 1.5 mile field test, improved only in the running group (l6%, p < 0.01; Table 2).
Bone mineral status Changes in bone mineral were identical whether expressed as BMC or as BMD, so only BMD results are given. Significant group-trial interactions were observed for lumbar spine BMD. A test of simple mean effects showed that pre-post differences in spine BMD for both the runners (0.013 + 0.019 g/cm’, mean + SD) and the weight trainers (0.015 + 0.018 g/cm’) were significant (p < 0.05). These pre-post differences were also significant (p < 0.05) when expressed as percentage changes, in which the lumbar spine BMD of the weight trainers and runners increased by 1.2 + 2.2 and 1.3 + 1.6070, respectively (Fig. 2). Responses in the exercise groups did not differ from each other but were significantly greater than those of the control subjects, in whom bone mineral did not change (p = 0.20; Fig. 2). N o measure of bone mineral at the proximal femur changed significantly in any group (Table 2). No significant relationship was observed between improvement in muscle strength or aerobic fitness and the change in bone mineral. Although group means for lumbar spine BMD increased significantly for both exercise groups, several women in each group lost bone (Fig. 2). No single factor or combination of factors predicted whether a woman gained or lost bone.
Activity profile Of the 31 women who completed the study, exercise in the previous year had been low in 11, moderate in 9, and high in 10. Among the subjects randomized to weight training, 6 were designated as low, 3 as moderate, and 3 as high. For runners, 3 were low, 5 were moderate, and 2 were high. For controls, 2 were low, 1 was moderate, and 5 were classified as highly active. There was no significant relationship between baseline BMD and previous exercise status, nor did previous exercise status predict the response to the training program. No significant differences in baseline activity profile were observed among the three groups (Table 3). No significant changes for any category of activity were observed
41.4(7.9)c 108.6(20.5)c 43.0(13.8)c 84.5(20.5)c 65.9(14.2)c 59.5 (11.O)d 39.5(6.3)c 14.6(4.7)
16.3(6.1)
15.5 (3.0)
28.2(10.6) 86.4(19.6) 34.1(6.3) 59.1 (16.6) 60.0(12.1) 53.6(15.1) 30.5 (9.0)
0.94(0.20) 0.75(0.12) 0.83(0.19)
0.97 (0.10) 0.81 (0.07) 0.88(0.13)
0.97(0.10) 0.80(0.08) 0.90 (0.12)
30.0(6.3) 85.5(14.2) 34.1 (8.8) 55.0 (9.4) 49.5(12.6) 54.1(9.4) 32.3(4.7)
1.11 (0.17)
162 (7.9) 60.2(14.8)
1.14(0.1 I)b
165 (4.7) 61.7(5.0)b
Pre
Post
13.0(2.7)d
29.1 (9.0) 87.7(19.6) 32.5(8.5) 65.0(15.1) 51.8(9.0) 59.5(15.1) 29.5(7.5)
0.94(0.19) 0.75(0.12) 0.83 (0.19)
1.13 (0.18)b
161 (10.3) 62.7(13.9)b
Runners (n = 10)
1.13 (0.12)
165 (4.5) 59.8(4.6)
Post
aMean (SD). Values indicated are significantly different from baseline (pre) values. bp < 0.05. cp < 0.001. dp < 0.01.
Height, cm Weight, kg Areal bone mineral density, g/cmZ Spine (L2-4) Proximal femur Neck Trochanter Ward's triangle Strength, kg Arm Shoulder Chest Leg Hip Back (deep) Back (sup) Aerobic fitness, minutes 1.5 mile walk/run
Pre
Weight trainers (n = 12)
TABLE2. BASELINE AND POSTEXERCISE MEASURES~
14.8(4.1)
33.2(10.3) 91.8(24.2) 35.2(11.6) 57.7(21.9) 49.5(10.3) 58.6(12.9) 35.0 (10.3)
0.93(0.15) 0.75(0.13) 0.82(0.18)
1.09(0.15)
167 (9.6) 61.2(6.2)
Pre
Post
~~~
15.4 (4.4)
35.5 (14.4) 91.8(32.1) 34.9(13.1) 57.7(24.4) 50.9 (1 1.6) 65.5 (12.9) 36.8(12.9)
0.93 (0.14) 0.75(0.12) 0.80(0.19)
1.08(0.13)
63.2(7.4)b
166 (7.0)
Controls (n = 8)
SNOW-HARTER ET AL.
T
i
t.
DISCUSSION
4%
I. Percentage changes in 1 RM muscle strength in the weight training group following 8 months of progressive resistance exercise. The error bars represent I SEM. Muscle strength for the arms, chest, shoulders, hips, back, and legs was recorded as summed values: arms = biceps curl + triceps press; chest = bench press; shoulders = military press + upright row; superficial back = lateral pull down; deep back = back extension; legs = leg extension + leg curl; hips = average hip abduction + average hip flexion. The term "average" refers to the mean score of the right and left sides. The increases were significant at the p < 0.01 (*) and p < 0.001 (**) levels compared to baseline values.
*
*
a
Tg + 1.396
4 ''
0
c
0
Weight
Run ne r s
in any group during the course of the study. Although there was an apparent increase in walking time and a decrease in recreational weight bearing exercise in the runners, these did not achieve significance (Table 3).
Controls
T ra i ne r s
FIG. 2. The filled circles represent the percentage change in lumbar spine BMD compared to baseline for individual subjects following 8 months of exercise training. The square notes the mean percentage change for each group, and the error bars represent 1 SD. The percentage change in the weight training and running groups was significantly different from zero (p < 0.05).
Relatively few intervention trials have examined the effect of resistance exercise on bone mineral. We have shown that 8 months of progressive training by either muscle strengthening or running exercise increased the lumbar spine mineral content and areal bone mineral density of young women. Dalsky and colleagues'1L)reported that postmenopausal participants in a 9 month program of walking, jogging, and light calisthenics increased lumbar spine BMD by 5.2%. The increase in lumbar spine BMD that we observed, about 1.2%. was relatively small compared to the results of Dalsky et a1.I") This may reflect the likelihood that the postmenopausal women were more sedentary than the young participants in the present study. Thus the change in daily skeletal loading brought about by the training may have been relatively greater for the older women. In interpreting this relatively small response to exercise intervention, it is important to know whether addition of a prescribed exercise routine leads to spontaneous decreases in other types of habitual activity. If this were the case, an overall increase in mechanical loading of the skeleton might not occur, and no change in bone mass would be expected. The activities of our subjects were self-reported by daily activity records during the course of this study. Within the constraints of self-reporting, no significant change in habitual activities was observed in any group. Thus, we conclude that the observed changes in bone mineral reflect increased mechanical loading specifically due to the exercise program. Our results appear to conflict with two exercise intervention trials in which weight lifting was the training mode. Gleeson et al.112)found that 1 year of weight training marginally increased lumbar spine density in a group of women. Although a significant difference in spine mineral was found between weight trainers and controls, the observed increase of 0.8% in bone mass over baseline values did not achieve significance. More recently, Rockwell et al.(I3)reported that women completing 1 year of weight training lost approximately 4% of lumbar spine mineral and questioned the safety of weight training for adult women. Several important differences distinguish both those studies from that reported here, so it may prove useful to compare the protocols in detail. The characteristics of all three studies are summarized in Table 4. Subjects in the 12 month study of Gleeson et aI.'l2)were older (24-46 years) than those in the present study. The majority of subjects were classified as sedentary, but 10 exercisers and 9 controls regularly participated in strenuous exercise before initiating the protocol. Whether this recreational activity continued throughout the study period is not stated. The training schedule consisted of upper and lower extremity exercises that did not specifically load the
767
EXERCISE TRAINING IN YOUNG WOMEN TABLE3. PROFILEOF DAILYACTIVITIES (H/DAY)~
Month I Weight trainers ( n = 12) Lying Sitting Standing Walking Recreation Weight bearing Other Runners (n = 10) Lying Sitting Standing Walking Recreation Weight bearing Other Controls ( n = 8) Lying Sitting Standing Walking Recreation Weight bearing Other aMean sions.
f SD.
8.38 9.68 2.98 1.62
Month 4
f 0.38 f 0.53 f 0.33 f
0.26
8.55 8.96 2.72 1.92
f f f f
Month 8
0.46 0.52 0.25 0.24
8.56 9.23 3.31 1.91
f f f f
0.36 0.56 0.42 0.26
0.22 f 0.08 0.65 f 0.15
0.41 f 0.15 0.67 f 0.19
0.16 f 0.07 0.51 f 0.14
8.74 9.84 2.89 1.56
8.94 9.56 3.64 1.91
8.94 9.32 3.04 2.00
f
0.33
f 0.36 f 0.39 f 0.19
0.22 f 0.10 0.35 f 0.07 7.42 9.41 3.89 1.29
f f f f
0.25 0.94 0.48 0.29
0.40 f 0.15 0.42 f 0.28
f f f f
0.78 0.74 0.57 0.21
f 0.42
f 0.75 f 0.55 f 0.29
0.06 f 0.05 0.45 f 0.23
0.13 f 0.06 0.25 f 0.09
8.10 9.99 3.56 1.24
8.86 9.53 3.97 1.41
f f f f
0.79 1.11 0.51 0.20
0.53 f 0.20 0.51 f 0.44
f 1.1
f 0.93 f 0.29 f 0.41
0.45 f 0.34 0.45 f 0.39
Scores in the recreation category d o not include study-related exercise ses-
TABLE4. COMPARISON OF RECENT STUDIESOF RESISTANCE EXERCISE
Gleeson et al.'L2)
DPA
72 (two groups)
30-40
nonrandom
Rockwell et al.1131
40 f 1.6
17 (two groups) nonrandom
DEXA
Current study
20
+
31 (three groups) random
DEXA
0.7
spine and emphasized relatively low loads and high repetitions. This schedule could result in a greater effect on muscle endurance than on dynamic strength. It is difficult to compare the reported strength gains to those we observed, since individual 1 RM scores were not given and the improvement was reported as upper and lower limb composites. The bone mineral measurements of Gleeson et al. ( I 2 ) were made by dual-photon absorptiometry, with a precision error of 2%. Although 34 participants completed
12 months, 3 times/week 8 exercises, 2 sets, 20 repeats 60% of 1 RM adjusted every 8 weeks 9 months, 2 times/week 8 exercises, 2 sets, 12 repeats 70% of 1 RM 9 months, 3 times/week 14 exercises, 3 sets, 8-12 repeats 70-85% of 1 RM adjusted every 6 weeks
weight training, the increase in bone mass was less than the precision error, and it is possible that statistical power was inadequate. Thus, despite its longer duration and greater number of subjects, a lower intensity of training and measurement precision may have adversely affected the outcome. In the study by Rockwell et aI.,lL3)participants were middle-aged (40 f 1.6 years) and relatively sedentary. However, several participants took part in regular exercise
768 before the start of the protocol. I t was stated that habitual activities continued throughout the study, but quantitative activity estimates were not recorded. The training schedule was based on 70% of 1 RM values, but it was not reported how the 1 RM values were obtained. Although the training protocol included exercises that specifically loaded the spine, it was isokinetic. The training stimulus was greater than that of Gleeson et al.; however, workouts were scheduled only twice each week, compared to three to four times per week in Gleeson et al. and in the present study. Although individual 1 RM values were not reported, it appears that strength improved substantially in all muscle groups. Bone mineral was measured by dual-energy x-ray absorptiometry with a precision error comparable to that reported here. The authors observed a reduction in serum calcium concentration and a rise in circulating parathyroid hormone and osteocalcin during the course of the study. Although these biochemical changes occurred in controls as well, they were more striking in the exercise group. The authors interpreted the data to suggest that bone turnover had increased in the exercising women. Although their study was conducted at a northern latitude during winter months, supplementation of participants with calcium and vitamin D argues against a seasonal change in vitamin D status as the basis for the observed biochemical alterations. Thus, weight training itself appears to have resulted in a 4% loss in lumbar spine mineral that cannot be attributed to any obvious confounding factors. It is possible that some of the women who jogged regularly before the weight training program decreased their running mileage sufficiently to experience a detraining effect, particularly in the early phases of weight training. It is also possible that the 9 month measurements reflect the initial resorption component of training-induced remodeling, before bone formation had gone to completion. In other words, the apparent loss could represent a temporary expansion of the “remodeling space.”(14) The results of the present study argue that a supervised program of progressive resistance training does not decrease bone mass and can actually increase lumbar spine mineral in young women. Like Gleeson et al.(lZL and Rockwell et al.,‘”’ we found no effect of exercise training on bone mineral at the proximal femur. This surprised us because both running and several of the strength training exercises specifically loaded the hip. It is possible that the greater abundance of cortical bone at this site may require more prolonged training for a detectable response, since remodeling occurs on bone surfaces and the surface to volume ratio of cortical bone is considerably less than that of trabecular bone. It may also be that habitual loading of the hip during such daily activities as standing and walking is so great that the increments in daily loading history produced by the exercise intervention were not important. We note that both Gleeson et al. and Rockwell et al. used nonrandomized group assignments. This strategy presumably reflects the view that compliance would be enhanced by permitting subjects to select their own program. In fact, attrition from both studies was about 35%. not very different from the 40% attrition rate in the present randomized trial, It is unlikely that the observed differ-
SNOW-HARTER ET AL. ences between our study and previous reports are simply a matter of randomization, but it is clear that randomization does not necessarily lead to excessive attrition or noncompliance. In view of these results, we conclude that young women can increase lumbar spine mineral through progressive exercise that involves either running or strength training. Although the increase in spine mineral achieved in this study was modest, it seems reasonable to predict that continuation of exercise for a longer period would lead to further increases in bone mass. We recognize that there may be constraints on the ultimate gains that can be achieved through exercise. This may be particularly true for populations, such as college students, whose habitual physical activity is already quite high. Maximal increases in bone mass will probably occur only when exercise continues to be progressive. If the training stimulus reaches a plateau, bone mass should not increase beyond the level necessary to accommodate the new loading conditions. These limitations are likely to be magnified when considering the role of exercise as a strategy to increase bone for the general population. In this study, as well as the others that have been d i ~ c u s s e d , ( ~ ~ exercise - ~ ~ . ’ ~ )training was carried out under expert scrutiny, with close attention given to safety and compliance. Even so, attrition rates above 30% were typical. It is likely that less tightly controlled programs would result in greater attrition, lower compliance, and smaller gains in bone mass. We note that previous exercise history did not predict baseline bone mineral density in our subjects. We used the exercise categories only to indicate the general activity level of our population and consider it inappropriate to conclude that habitual activity level does not influence bone mass in young women. The surveys on which exercise designations were made referred to all recreational activity without regard for intensity level or activities that preferentially load the skeleton. They also did not include nonrecreational activity, such as stair climbing, lifting, or other incidental types of skeletal loading. Although a primary hypothesis for this study stated that weight training would provide a superior osteogenic stimulus, the results indicate equivalence of the two exercise modes when carried out at the intensity and relatively short duration used in this study. From the bone remodeling theories(*,s’we predicted that the skeletal response would be greater with strength training, a high load-low repetition activity, than with running, a high repetition-low load activity. The results of this study d o not appear to support this prediction. However, this interpretation must be tempered by several factors. The number of load repetitions from running may have been so great that an osteogenic stimulus equivalent to that from the weight training was actually provided. Moreover, the theoretical model defines an equilibrium relationship and is not capable of describing the dynamic response of bone to a change in mechanical loading. Although hypertrophy of muscle occurs within several weeks of the onset of strength training, bone may react more slowly, requiring a longer training period to differentiate the ultimate effects of different training modalities.
EXERCISE TRAINING IN YOUNG WOMEN
769
ACKNOWLEDGMENTS The authors are pleased to express thanks to Susan Charette and Pamela Weinstein for their dedication and hard work in supervising the training and testing of our subjects, and to Marc Graeber, M.D. for his continuing enthusiastic support of our laboratory. The cooperation of Betsy Weeks of the Department of Athletics, Physical Education, and Recreation, Stanford University in providing access to the Roble Gymnasium was critical to the success of this program. We acknowledge the generous contribution of Rorer, Inc. for their gift of Calel-D for our subjects. We thank Byron W. Brown, Jr. for statistical advice. Finally, we gratefully acknowledge the persistence and good nature of our participants. This study was supported by NIH Grant AR 38941 and by the Research Service of the Department of Veterans Affairs.
7.
8. 9. 10.
11.
12.
13.
REFERENCES 1 . Marcus R, Carter DR 1988 The role of physical activity in bone mass regulation. Adv Sports Med Fitness 1:63-82. 2. Carter DR, Fyhrie DP, Whalen RT 1987 Trabecular bone density and loading history: Regulation of connective tissue biology by mechanical energy. J Biomech 20:785-794. 3. Frost H 1987 The mechanostat: A proposed pathogenic mechanism of osteoporosis and the bone mass effects of mechanical and nonmechanical agents. Bone Miner 2:73-86. 4. Rubin CT, Lanyon LE 1984 Regulation of bone formation by applied dynamic loads. J Bone Joint Surg [Am 66~397-402. 5 . Whalen RT, Carter DR, Steele DR 1988 Influence of physical activity on the regulation of bone density. J Biomech 21:825-837. 6. Heinrich C, Going S, Pamenter R, Perry S, Boyden T , Lohman T 1990 Bone mineral content of cyclically menstruating
14.
15.
female resistance and endurance trained athletes. Med Sci Sports Exerc 2 2 3 8 - 5 6 3 . Snow-Harter C , Bouxsein M, Lewis B, Charette S, Weinstein P, Marcus R 1990 Muscle strength as a predictor of bone mineral density in young women. J Bone Miner Res 5589595. Nieman D 1986 The Sports Medicine Fitness Course. Bull Publishing, Palo Alto, California. Heyward V 1984 Designs for Fitness. Burgess Publishing, Minneapolis, Minnesota. AAHPERD 1984 Technical Manual: Health Related Physical Fitness. American Alliance for Health, Physical Education, Recreation, and Dance, Reston, California. Dalsky G P , Stocke KS, Ehsani AA, Slatopolsky E, Lee WC, Birge SJ 1988 Weight-bearing exercise training and lumbar bone mineral content in postmenopausal women. Ann Intern Med 1082324-828. Gleeson PB, Protas E J , LeBlanc AD, Schneider VS, Evans H J 1990 Effects of weight lifting on bone mineral density in premenopausal women. J Bone Miner Res 5153-158. Rockwell J , Sorensen A, Baker S , Leahey D, Stock J , Michaels J, Baran D 1990 Weight training decreases vertebral bone density in premenopausal women: A prospective study. J Clin Endocrinol Metab 71:988-993. Jaworski Z 1976 Parameters and indices of bone resorption. In: Meunier P (ed) Bone Histomorphometry, 2nd International Workshop. Armour Montague, Paris, pp. 193-200. Cavanaugh DJ, Cann CE 1988 Brisk walking does not stop bone loss in postmenopausal women. Bone 9:201-204.
Address reprint requests to: Robert Marcus, M.D. GRECC, 182-B VA Medical Center Palo Alto, CA 94304 Received for publication June 19, 1991; in revised form January 22, 1992; accepted February 7, 1992.
Effects of Resistance and Endurance Exercise on Bone Mineral Status of Young Women: A Randomized Exercise Intervention Trial CHRISTINE SNOW-HARTER,* MARY L. BOUXSEIN, BARBARA T. LEWIS, DENNIS R. CARTER, and ROBERT MARCUS
ABSTRACT A substantial body of cross-sectional data and a smaller number of intervention trials generally justify optimism that regular physical activity benefits the skeleton. We conducted an 8 month controlled exercise trial in a group of healthy college women (mean age = 19.9 years) who were randomly assigned to a control group or to progressive training in jogging or weight lifting. We measured the following variables: bone mineral density (BMD) of the spine (L2-4) and right proximal femur using dual-energy x-ray absorptiometry, dynamic muscle strength using the 1-RM method, and endurance performance using the 1.5 mile walk/run field test. A total of 31 women completed the 8 month study. For women completing the study, compliance, defined as the percentage of workout sessions attended, was 97% for the runners (range 90-100%) and 92% (range 88-lOoV0) for the weight trainers. Body weight increased by approximately 2 kg in all groups 0, < 0.05). Weight training was associated with significant increases (p < 0.01) in muscle strength in all muscle groups. Improvement ranged from 10% for the deep back to 54% for the leg. No significant changes in strength scores were observed in the control or running groups. Aerobic performance improved only in the running group (l6V0, p < 0.01). Lumbar BMD increased (p < 0.05) in both runners (1.3 f 1.6%) and weight trainers (1.2 f 1.8%). These results did not differ from each other but were both significantiy greater than results in control subjects, in whom bone mineral did not change. N o measure of bone mineral at the proximal femur changed significantly in any group. These results demonstrate that 8 months of supervised progressive training in either running or resistance exercise modestly increases lumbar spine mineral in young women.
INTRODUCTION justifies optimism for the concept that regular physical exercise benefits the skeleton. Exercise prescriptions may be defined according to type, intensity, duration, and frequency, but we do not yet understand the details of a program that would be optimal for bone.'') Bone adapts to the loads applied to it. This adaptation includes an increase in density when mechanical loading increases and a loss of bone density when customary loads are removed. This concept has been expanded
C
ONSIDERABLE EVIDENCE
to a general theory of bone remodeling in which mechanical loading is a primary regulator. ('-') The mechanisms by which mechanical forces alter bone mass await clarification, but Carter et al.(2)proposed that the apparent density of a bone region is determined by its customary loading history. To expand this concept, Whalen et al.'s' proposed a mathematical model that relates bone mass at equilibrium to its typical loading experience. This daily history is defined as the sum of a series of loading events, each characterized by a load magnitude and number of cycles. The studies of Whalen et aI.(') suggest that the ef-
Musculoskeletal Research Laboratory, GRECC, Department of Veterans Affairs Medical Center, Palo Alto, and the Departments of Medicine and Mechanical Engineering, Stanford University, Stanford, California. *Current address: Department of Exercise and Sports Science, Oregon State University, Corvallis.
761
SNOW-HARTER ET AL.
762 fect of load magnitude should outweigh that of cycle number. If this formulation is correct, it would be a reasonable hypothesis that muscle strength training, an activity of high loads and few repetitions, would have a greater effect on bone mass than endurance exercise, a high repetitionlow magnitude activity. In this regard, Heinrich et a1.‘61 found that axial and peripheral bone mass were higher in young women who participated in weight lifting exercise than in endurance-trained women of similar age. We conducted a controlled exercise intervention trial in a group of healthy college women who were randomly assigned to a control group or to a program of progressive training in jogging or weight lifting. This study addressed the following two hypotheses: (1) exercise training increases bone mineral content of the lumbar spine and proximal femur, and (2) changes in spine and proximal femur mineral content are greater with weight training than with jogging. In this paper we present the experimental results of this study.
MATERIALS AND METHODS Subjects A total of 52 women, age 19.9 f 0.7 years (mean f standard error of the mean, SEM), were recruited from the student body of Stanford University. Volunteers who were in good health with regular menstrual cycles (8-12 periods per year) for at least 3 years before the study and who were without chronic physical problems that would limit participation in an exercise program were invited to participate in this study. Competitive athletes were not invited to enroll, and no subject belonged to an athletic team. On acceptance into the study, subjects were randomly assigned to control, weight training, or running groups. All baseline measurements were completed before initiating the exercise program. Subjects participated in the exercise sessions, completed activity records and exercise logs, and attended group meetings. We defined satisfactory participation in the exercise sessions as attending at least 85% of all sessions and completing >90% of the designated workout at each session attended. All testing and training sessions were conducted under supervision of the research staff. The protocol was approved by the Human Subjects Committee, Stanford University, and all subjects gave written consent.
Bone mineral evaluation Bone mineral content (BMC, g) and areal bone mineral density (BMD, g/cm’) of the lumbar spine (L2-4) and right proximal femur were estimated by dual-energy x-ray absorptiometry (Hologic QDR 1O00, Waltham, MA) on entry and following 8 months of training. The femoral neck, trochanter, and Ward‘s triangle regions of the hip were analyzed separately. In our laboratory, the following precision errors for BMD (coefficients of variation for multiple determinations) have been determined for a group of 20 healthy young adult women: (1) lumbar spine = 0.5%, (2) femoral neck and Ward’s triangle = 0.9%, and (3) trochanter = 0.75%. Bone phantoms (Hologic Corp., Waltham, MA) were scanned each day throughout the study and gave readings that were consistently within 0.7% of nominal values. N o evidence for machine drift appeared during this study.
Muscle strength assessment After formal instruction in the use of weight machines, all subjects underwent an initial assessment of muscle strength. A total of 13 exercises were selected to assess the strength of the arms, chest, shoulder, back, hips, and legs (Table 1). Biceps curl, triceps extension, bench press, military press, upright row, and knee extension and flexion were performed on a Universal Multi-station weight machine (Universal Gym Equipment, Inc., Cedar Rapids, IA). Hip adduction, abduction, extension, and flexion were executed on a Universal Total Hip Trainer model 993494 (Universal Gym Equipment, Inc., Cedar Rapids, IA). Back extension was tested o n the Nautilus Low Back Machine (Nautilus Sports/Medicine Industries, Dallas, TX). Dynamic strength of all major muscle groups was assessed using the one repetition maximum (1 RM) method, as we have validated and reported.‘’) One RM is the maximum weight a subject can lift one time with acceptable form. “Acceptable form” means that performance of the exercise isolated primarily the muscle groups being tested with minimal recruitment of other major muscle groups. When the criterion for acceptable form was not satisfied, the last recorded successful weight was taken for the strength score. The details of each exercise have been de-
TABLE1. EXERCISES USED FOR STRENGTH TESTSAND
Exercise Biceps curl Triceps press Upright row Bench press Sit-ups Military press Lateral pull down
BY
WEIGHT TRAINERS IN WORKOUTSa
Primary muscles
Exercise
Primary muscles
Biceps brachii, brachialis Triceps brachii Deltoids, trapezius Pectoralis major, triceps Rectus abdominus Deltoids, triceps Latissimus dorsi
Knee flexion Back extension Knee extension Hip adduction Hip abduction Hip extension Hip flexion
Hamstrings group Erector spinae Quadriceps group Adductor magnus Gluteus medius Gluteus maximus lliopsoas
alncluded are primary muscles activated to perform exercise.
763
EXERCISE TRAINING IN YOUNG WOMEN scribed previously. (’) Before each strength-testing session, subjects received detailed instructions and performed each exercise two times at very low resistance to enhance familiarization and warm-up. Each 1-RM test began at a weight near the anticipated maximum to minimize fatigue resulting from repetition. In most cases subjects required no more than three repetitions to reach maximum. All tests were repeated with weight increases of 2.5, 5 , and 10 pounds, depending upon the exercise, until the subject was not able to lift additional weight with acceptable form. A 20 s rest period was given between repetitions, with at least 1 minute of rest between exercises. Strength tests were administered on 2 nonconsecutive days, with hip and back strength recorded on 1 day and arm, shoulder, chest, superficial back, and leg strength assessed on the second test date. In our laboratory, the coefficient of variation for repeated measures is 2-6% for all strength tests in young women. These measurements served as baseline values for subjects in the control and running groups as well as for the weight trainers. Muscle strength for the arms, chest, shoulders, hips, back, and legs was recorded as summed values obtained from the 1-RM tests as follows: arms = biceps + triceps press; chest = bench press; shoulders = military press + upright row; superficial back = lateral pull down; deep back = back extension; legs = leg extension + leg curl; hips = average hip abduction + average hip flexion. The term “average” refers to the mean score of the right and left sides. Hip extension and adduction are not included in the expression of hip strength because many subjects had not reached maximum with the entire stack of weight plates.
Cardiorespiratory fitness testing The 1.5 mile walklrun field test was conducted to assess endurance performance.“’ In this simple test, subjects are instructed to travel 1.5 miles on foot as quickly as possible. Any combination of running and walking is permitted. All subjects performed this test at the Stanford University track at the same time of day after a standardized warmup period. Results are scored as the time required to complete the 1.5 mile task. Performance on this test correlates strongly with measures of aerobic fitness obtained from Vozrnax In a pilot study involving 10 young women we found the precision error of this test to be 7%.
Activity assessments O n entry, subjects completed a questionnaire detailing their exercise history during the previous year. Subjects were asked to itemize all recreational activities in which they had participated. They listed the number of hours per week and the number of months per year that they participated in each listed activity. From this questionnaire, the average number of hours of exercise per week was calculated and the women were grouped into three categories: low 5 1 h/week, moderate = 2-3 h/week, and high > 3 h/week. Average daily activity profiles were developed for each
participant from daily activity logs that were collected during the study (October, January, and May). Subjects recorded their activities for 7 consecutive days, registering for each day the number of hours that were spent lying, sitting, standing, walking, and participating in both weight-bearing and non-weight-bearing recreational activities other than those directly related to the training sessions. Mean daily values were calculated for each activity. Subjects received instruction on record maintenance. On submission, records that did not add up to 23-25 h of activity were discarded and subjects were asked to complete records for additional days.
Exercise training protocols This study was designed to correspond to a 35 week school year. Exercise training commenced in early October and finished during the last week in May. Muscle strength, bone mineral, and aerobic fitness assessments were made within 1 week before beginning and following the end of training. Training was interrupted from weeks 12 to 14 during the winter holiday and week 26 during the spring vacation. Although some runners were able to continue exercise, few weight trainers had access to the gymnasium at these times. For women in the weight training group, no reduction in training load occurred following these breaks in training. Weight Training: Subjects performed a circuit of 14 exercises three times per week under supervision. The exercises were identical to those enumerated in Table 1 and have been previously described in detail.(’) At each workout session, the subjects performed all exercises for three sets of 8-12 repetitions each. Training was progressive, with an increase in load of not more than 10% per week. Training began at 6 5 7 0 % of baseline 1 RM values and reached target levels of 75% of the initial 1 RM within 3 months. Strength testing was conducted every 6 weeks throughout the 8 month training period, and training levels were adjusted accordingly. During the last 2 months of the study, training was conducted at 85% of the current 1 RM values. The back extension exercise was the sole exception to this training schedule, since a Nautilus back machine was not available at the training facility. Therefore, to load the deep back musculature, back extensions were performed to exhaustion using lower weights.
Running: Before training was initiated all subjects attended a session in which the following topics were discussed: running form and biomechanics, foot placement and stride, injury prevention, and shoe selection. Subjects ran three times or more each week. At least two-thirds of running sessions were carried out under supervision. Each session began with a warm-up period that included 3-5 minutes of slow jogging followed by stretching and finished with several minutes of walking and stretching. Subjects were instructed to run at a speed sufficient to elevate the heart rate to 70-80% of age-predicted maxima. Running schedules were designed for each subject based on her 1.5 mile walk/run time, and mileage increases were re-
SNOW-HARTER ET AL.
764 stricted to 10% per week. The progression of average weekly mileage was as follows (mean + standard deviation, SD): months 1 and 2 = 4.2 + 2.3 miles, months 3 and 4 = 5.6 + 4.3 miles, months 5 and 6 = 1.9 + 5.1 miles, and months 7 and 8 = 10.5 f 4.9 miles.
Control Group: The control subjects did not change physical activity patterns from baseline. Maintenance of recreational activity at baseline levels was encouraged.
Miscellaneous Meetings were held quarterly for all subjects. Exercise logs were maintained for each workout. For runners these logs included the distance run and the pace. For weight trainers, the number of repetitions and weight lifted were recorded for all exercises. All subjects were issued a daily 500 mg supplement of calcium as carbonate (Calel-D, Rorer Pharmaceuticals, Inc., Fort Washington, PA) for the duration of the study. Incentives were offered to exercising subjects throughout the study period. Awards were given to subjects achieving 80% of stated goals, with additional prizes for exceptional compliance.
Data management and analysis Data were analyzed using the Statview I1 software package (Abacus Concepts, Inc., Berkeley, CA). Analyses included standard descriptive statistics, paired t-tests, and analysis of variance (ANOVA). Changes over time in muscle strength were analyzed as arithmetic before-after differences as well as by percentage differences. Changes in BMD were analyzed by 3 x 2 ANOVA with repeated measures on the trial factor. T h e p values reported for paired ttests were two-tailed. Since bone mineral at the spine was the primary study end point, an a value of 0.05 was taken as the criterion for significance. For other bone sites and for muscle strength changes, (Y = 0.01 was chosen to adjust for multiple comparisons.
RESULTS Subject characteristics (means + SD) on entry are presented in Table 2. No women were obese, as BMI (weight +- height’) was below 30 in all cases. All baseline variables were normally distributed and did not differ significantly among groups, indicating that randomization was effective. A total of 31 women completed the 8 month study, representing an attrition of 40%. Subjects who left the study did so for academic and personal reasons. No subject stopped participation because of injury. One runner developed sustained amenorrhea during the study period. Although she completed the study, we excluded her data from analysis. For women completing the study, compliance, defined as the percentage of workout sessions attended, was 97% for the runners (range 90-100%) and 92% (range 88100%) for the weight trainers. The training programs were well tolerated by all participants. No subject developed
overuse or other injury of sufficient magnitude to require modification or interruption of the training schedule.
Body weight, muscle strength, and endurance performance Pre- and posttraining values are presented in Table 2. Body weight increased by approximately 2 kg in all groups (p < 0.05). Weight training was associated with significant increases (p < 0.01) in muscle strength in all muscle groups (Table 2 and Fig. 1). The magnitude of improvement ranged from 10% for the deep back muscles to 54% for the leg (knee flexors and extensors). No significant changes in strength scores were observed in the control or running groups. Endurance performance, as measured by the 1.5 mile field test, improved only in the running group (l6%, p < 0.01; Table 2).
Bone mineral status Changes in bone mineral were identical whether expressed as BMC or as BMD, so only BMD results are given. Significant group-trial interactions were observed for lumbar spine BMD. A test of simple mean effects showed that pre-post differences in spine BMD for both the runners (0.013 + 0.019 g/cm’, mean + SD) and the weight trainers (0.015 + 0.018 g/cm’) were significant (p < 0.05). These pre-post differences were also significant (p < 0.05) when expressed as percentage changes, in which the lumbar spine BMD of the weight trainers and runners increased by 1.2 + 2.2 and 1.3 + 1.6070, respectively (Fig. 2). Responses in the exercise groups did not differ from each other but were significantly greater than those of the control subjects, in whom bone mineral did not change (p = 0.20; Fig. 2). N o measure of bone mineral at the proximal femur changed significantly in any group (Table 2). No significant relationship was observed between improvement in muscle strength or aerobic fitness and the change in bone mineral. Although group means for lumbar spine BMD increased significantly for both exercise groups, several women in each group lost bone (Fig. 2). No single factor or combination of factors predicted whether a woman gained or lost bone.
Activity profile Of the 31 women who completed the study, exercise in the previous year had been low in 11, moderate in 9, and high in 10. Among the subjects randomized to weight training, 6 were designated as low, 3 as moderate, and 3 as high. For runners, 3 were low, 5 were moderate, and 2 were high. For controls, 2 were low, 1 was moderate, and 5 were classified as highly active. There was no significant relationship between baseline BMD and previous exercise status, nor did previous exercise status predict the response to the training program. No significant differences in baseline activity profile were observed among the three groups (Table 3). No significant changes for any category of activity were observed
41.4(7.9)c 108.6(20.5)c 43.0(13.8)c 84.5(20.5)c 65.9(14.2)c 59.5 (11.O)d 39.5(6.3)c 14.6(4.7)
16.3(6.1)
15.5 (3.0)
28.2(10.6) 86.4(19.6) 34.1(6.3) 59.1 (16.6) 60.0(12.1) 53.6(15.1) 30.5 (9.0)
0.94(0.20) 0.75(0.12) 0.83(0.19)
0.97 (0.10) 0.81 (0.07) 0.88(0.13)
0.97(0.10) 0.80(0.08) 0.90 (0.12)
30.0(6.3) 85.5(14.2) 34.1 (8.8) 55.0 (9.4) 49.5(12.6) 54.1(9.4) 32.3(4.7)
1.11 (0.17)
162 (7.9) 60.2(14.8)
1.14(0.1 I)b
165 (4.7) 61.7(5.0)b
Pre
Post
13.0(2.7)d
29.1 (9.0) 87.7(19.6) 32.5(8.5) 65.0(15.1) 51.8(9.0) 59.5(15.1) 29.5(7.5)
0.94(0.19) 0.75(0.12) 0.83 (0.19)
1.13 (0.18)b
161 (10.3) 62.7(13.9)b
Runners (n = 10)
1.13 (0.12)
165 (4.5) 59.8(4.6)
Post
aMean (SD). Values indicated are significantly different from baseline (pre) values. bp < 0.05. cp < 0.001. dp < 0.01.
Height, cm Weight, kg Areal bone mineral density, g/cmZ Spine (L2-4) Proximal femur Neck Trochanter Ward's triangle Strength, kg Arm Shoulder Chest Leg Hip Back (deep) Back (sup) Aerobic fitness, minutes 1.5 mile walk/run
Pre
Weight trainers (n = 12)
TABLE2. BASELINE AND POSTEXERCISE MEASURES~
14.8(4.1)
33.2(10.3) 91.8(24.2) 35.2(11.6) 57.7(21.9) 49.5(10.3) 58.6(12.9) 35.0 (10.3)
0.93(0.15) 0.75(0.13) 0.82(0.18)
1.09(0.15)
167 (9.6) 61.2(6.2)
Pre
Post
~~~
15.4 (4.4)
35.5 (14.4) 91.8(32.1) 34.9(13.1) 57.7(24.4) 50.9 (1 1.6) 65.5 (12.9) 36.8(12.9)
0.93 (0.14) 0.75(0.12) 0.80(0.19)
1.08(0.13)
63.2(7.4)b
166 (7.0)
Controls (n = 8)
SNOW-HARTER ET AL.
T
i
t.
DISCUSSION
4%
I. Percentage changes in 1 RM muscle strength in the weight training group following 8 months of progressive resistance exercise. The error bars represent I SEM. Muscle strength for the arms, chest, shoulders, hips, back, and legs was recorded as summed values: arms = biceps curl + triceps press; chest = bench press; shoulders = military press + upright row; superficial back = lateral pull down; deep back = back extension; legs = leg extension + leg curl; hips = average hip abduction + average hip flexion. The term "average" refers to the mean score of the right and left sides. The increases were significant at the p < 0.01 (*) and p < 0.001 (**) levels compared to baseline values.
*
*
a
Tg + 1.396
4 ''
0
c
0
Weight
Run ne r s
in any group during the course of the study. Although there was an apparent increase in walking time and a decrease in recreational weight bearing exercise in the runners, these did not achieve significance (Table 3).
Controls
T ra i ne r s
FIG. 2. The filled circles represent the percentage change in lumbar spine BMD compared to baseline for individual subjects following 8 months of exercise training. The square notes the mean percentage change for each group, and the error bars represent 1 SD. The percentage change in the weight training and running groups was significantly different from zero (p < 0.05).
Relatively few intervention trials have examined the effect of resistance exercise on bone mineral. We have shown that 8 months of progressive training by either muscle strengthening or running exercise increased the lumbar spine mineral content and areal bone mineral density of young women. Dalsky and colleagues'1L)reported that postmenopausal participants in a 9 month program of walking, jogging, and light calisthenics increased lumbar spine BMD by 5.2%. The increase in lumbar spine BMD that we observed, about 1.2%. was relatively small compared to the results of Dalsky et a1.I") This may reflect the likelihood that the postmenopausal women were more sedentary than the young participants in the present study. Thus the change in daily skeletal loading brought about by the training may have been relatively greater for the older women. In interpreting this relatively small response to exercise intervention, it is important to know whether addition of a prescribed exercise routine leads to spontaneous decreases in other types of habitual activity. If this were the case, an overall increase in mechanical loading of the skeleton might not occur, and no change in bone mass would be expected. The activities of our subjects were self-reported by daily activity records during the course of this study. Within the constraints of self-reporting, no significant change in habitual activities was observed in any group. Thus, we conclude that the observed changes in bone mineral reflect increased mechanical loading specifically due to the exercise program. Our results appear to conflict with two exercise intervention trials in which weight lifting was the training mode. Gleeson et al.112)found that 1 year of weight training marginally increased lumbar spine density in a group of women. Although a significant difference in spine mineral was found between weight trainers and controls, the observed increase of 0.8% in bone mass over baseline values did not achieve significance. More recently, Rockwell et al.(I3)reported that women completing 1 year of weight training lost approximately 4% of lumbar spine mineral and questioned the safety of weight training for adult women. Several important differences distinguish both those studies from that reported here, so it may prove useful to compare the protocols in detail. The characteristics of all three studies are summarized in Table 4. Subjects in the 12 month study of Gleeson et aI.'l2)were older (24-46 years) than those in the present study. The majority of subjects were classified as sedentary, but 10 exercisers and 9 controls regularly participated in strenuous exercise before initiating the protocol. Whether this recreational activity continued throughout the study period is not stated. The training schedule consisted of upper and lower extremity exercises that did not specifically load the
767
EXERCISE TRAINING IN YOUNG WOMEN TABLE3. PROFILEOF DAILYACTIVITIES (H/DAY)~
Month I Weight trainers ( n = 12) Lying Sitting Standing Walking Recreation Weight bearing Other Runners (n = 10) Lying Sitting Standing Walking Recreation Weight bearing Other Controls ( n = 8) Lying Sitting Standing Walking Recreation Weight bearing Other aMean sions.
f SD.
8.38 9.68 2.98 1.62
Month 4
f 0.38 f 0.53 f 0.33 f
0.26
8.55 8.96 2.72 1.92
f f f f
Month 8
0.46 0.52 0.25 0.24
8.56 9.23 3.31 1.91
f f f f
0.36 0.56 0.42 0.26
0.22 f 0.08 0.65 f 0.15
0.41 f 0.15 0.67 f 0.19
0.16 f 0.07 0.51 f 0.14
8.74 9.84 2.89 1.56
8.94 9.56 3.64 1.91
8.94 9.32 3.04 2.00
f
0.33
f 0.36 f 0.39 f 0.19
0.22 f 0.10 0.35 f 0.07 7.42 9.41 3.89 1.29
f f f f
0.25 0.94 0.48 0.29
0.40 f 0.15 0.42 f 0.28
f f f f
0.78 0.74 0.57 0.21
f 0.42
f 0.75 f 0.55 f 0.29
0.06 f 0.05 0.45 f 0.23
0.13 f 0.06 0.25 f 0.09
8.10 9.99 3.56 1.24
8.86 9.53 3.97 1.41
f f f f
0.79 1.11 0.51 0.20
0.53 f 0.20 0.51 f 0.44
f 1.1
f 0.93 f 0.29 f 0.41
0.45 f 0.34 0.45 f 0.39
Scores in the recreation category d o not include study-related exercise ses-
TABLE4. COMPARISON OF RECENT STUDIESOF RESISTANCE EXERCISE
Gleeson et al.'L2)
DPA
72 (two groups)
30-40
nonrandom
Rockwell et al.1131
40 f 1.6
17 (two groups) nonrandom
DEXA
Current study
20
+
31 (three groups) random
DEXA
0.7
spine and emphasized relatively low loads and high repetitions. This schedule could result in a greater effect on muscle endurance than on dynamic strength. It is difficult to compare the reported strength gains to those we observed, since individual 1 RM scores were not given and the improvement was reported as upper and lower limb composites. The bone mineral measurements of Gleeson et al. ( I 2 ) were made by dual-photon absorptiometry, with a precision error of 2%. Although 34 participants completed
12 months, 3 times/week 8 exercises, 2 sets, 20 repeats 60% of 1 RM adjusted every 8 weeks 9 months, 2 times/week 8 exercises, 2 sets, 12 repeats 70% of 1 RM 9 months, 3 times/week 14 exercises, 3 sets, 8-12 repeats 70-85% of 1 RM adjusted every 6 weeks
weight training, the increase in bone mass was less than the precision error, and it is possible that statistical power was inadequate. Thus, despite its longer duration and greater number of subjects, a lower intensity of training and measurement precision may have adversely affected the outcome. In the study by Rockwell et aI.,lL3)participants were middle-aged (40 f 1.6 years) and relatively sedentary. However, several participants took part in regular exercise
768 before the start of the protocol. I t was stated that habitual activities continued throughout the study, but quantitative activity estimates were not recorded. The training schedule was based on 70% of 1 RM values, but it was not reported how the 1 RM values were obtained. Although the training protocol included exercises that specifically loaded the spine, it was isokinetic. The training stimulus was greater than that of Gleeson et al.; however, workouts were scheduled only twice each week, compared to three to four times per week in Gleeson et al. and in the present study. Although individual 1 RM values were not reported, it appears that strength improved substantially in all muscle groups. Bone mineral was measured by dual-energy x-ray absorptiometry with a precision error comparable to that reported here. The authors observed a reduction in serum calcium concentration and a rise in circulating parathyroid hormone and osteocalcin during the course of the study. Although these biochemical changes occurred in controls as well, they were more striking in the exercise group. The authors interpreted the data to suggest that bone turnover had increased in the exercising women. Although their study was conducted at a northern latitude during winter months, supplementation of participants with calcium and vitamin D argues against a seasonal change in vitamin D status as the basis for the observed biochemical alterations. Thus, weight training itself appears to have resulted in a 4% loss in lumbar spine mineral that cannot be attributed to any obvious confounding factors. It is possible that some of the women who jogged regularly before the weight training program decreased their running mileage sufficiently to experience a detraining effect, particularly in the early phases of weight training. It is also possible that the 9 month measurements reflect the initial resorption component of training-induced remodeling, before bone formation had gone to completion. In other words, the apparent loss could represent a temporary expansion of the “remodeling space.”(14) The results of the present study argue that a supervised program of progressive resistance training does not decrease bone mass and can actually increase lumbar spine mineral in young women. Like Gleeson et al.(lZL and Rockwell et al.,‘”’ we found no effect of exercise training on bone mineral at the proximal femur. This surprised us because both running and several of the strength training exercises specifically loaded the hip. It is possible that the greater abundance of cortical bone at this site may require more prolonged training for a detectable response, since remodeling occurs on bone surfaces and the surface to volume ratio of cortical bone is considerably less than that of trabecular bone. It may also be that habitual loading of the hip during such daily activities as standing and walking is so great that the increments in daily loading history produced by the exercise intervention were not important. We note that both Gleeson et al. and Rockwell et al. used nonrandomized group assignments. This strategy presumably reflects the view that compliance would be enhanced by permitting subjects to select their own program. In fact, attrition from both studies was about 35%. not very different from the 40% attrition rate in the present randomized trial, It is unlikely that the observed differ-
SNOW-HARTER ET AL. ences between our study and previous reports are simply a matter of randomization, but it is clear that randomization does not necessarily lead to excessive attrition or noncompliance. In view of these results, we conclude that young women can increase lumbar spine mineral through progressive exercise that involves either running or strength training. Although the increase in spine mineral achieved in this study was modest, it seems reasonable to predict that continuation of exercise for a longer period would lead to further increases in bone mass. We recognize that there may be constraints on the ultimate gains that can be achieved through exercise. This may be particularly true for populations, such as college students, whose habitual physical activity is already quite high. Maximal increases in bone mass will probably occur only when exercise continues to be progressive. If the training stimulus reaches a plateau, bone mass should not increase beyond the level necessary to accommodate the new loading conditions. These limitations are likely to be magnified when considering the role of exercise as a strategy to increase bone for the general population. In this study, as well as the others that have been d i ~ c u s s e d , ( ~ ~ exercise - ~ ~ . ’ ~ )training was carried out under expert scrutiny, with close attention given to safety and compliance. Even so, attrition rates above 30% were typical. It is likely that less tightly controlled programs would result in greater attrition, lower compliance, and smaller gains in bone mass. We note that previous exercise history did not predict baseline bone mineral density in our subjects. We used the exercise categories only to indicate the general activity level of our population and consider it inappropriate to conclude that habitual activity level does not influence bone mass in young women. The surveys on which exercise designations were made referred to all recreational activity without regard for intensity level or activities that preferentially load the skeleton. They also did not include nonrecreational activity, such as stair climbing, lifting, or other incidental types of skeletal loading. Although a primary hypothesis for this study stated that weight training would provide a superior osteogenic stimulus, the results indicate equivalence of the two exercise modes when carried out at the intensity and relatively short duration used in this study. From the bone remodeling theories(*,s’we predicted that the skeletal response would be greater with strength training, a high load-low repetition activity, than with running, a high repetition-low load activity. The results of this study d o not appear to support this prediction. However, this interpretation must be tempered by several factors. The number of load repetitions from running may have been so great that an osteogenic stimulus equivalent to that from the weight training was actually provided. Moreover, the theoretical model defines an equilibrium relationship and is not capable of describing the dynamic response of bone to a change in mechanical loading. Although hypertrophy of muscle occurs within several weeks of the onset of strength training, bone may react more slowly, requiring a longer training period to differentiate the ultimate effects of different training modalities.
EXERCISE TRAINING IN YOUNG WOMEN
769
ACKNOWLEDGMENTS The authors are pleased to express thanks to Susan Charette and Pamela Weinstein for their dedication and hard work in supervising the training and testing of our subjects, and to Marc Graeber, M.D. for his continuing enthusiastic support of our laboratory. The cooperation of Betsy Weeks of the Department of Athletics, Physical Education, and Recreation, Stanford University in providing access to the Roble Gymnasium was critical to the success of this program. We acknowledge the generous contribution of Rorer, Inc. for their gift of Calel-D for our subjects. We thank Byron W. Brown, Jr. for statistical advice. Finally, we gratefully acknowledge the persistence and good nature of our participants. This study was supported by NIH Grant AR 38941 and by the Research Service of the Department of Veterans Affairs.
7.
8. 9. 10.
11.
12.
13.
REFERENCES 1 . Marcus R, Carter DR 1988 The role of physical activity in bone mass regulation. Adv Sports Med Fitness 1:63-82. 2. Carter DR, Fyhrie DP, Whalen RT 1987 Trabecular bone density and loading history: Regulation of connective tissue biology by mechanical energy. J Biomech 20:785-794. 3. Frost H 1987 The mechanostat: A proposed pathogenic mechanism of osteoporosis and the bone mass effects of mechanical and nonmechanical agents. Bone Miner 2:73-86. 4. Rubin CT, Lanyon LE 1984 Regulation of bone formation by applied dynamic loads. J Bone Joint Surg [Am 66~397-402. 5 . Whalen RT, Carter DR, Steele DR 1988 Influence of physical activity on the regulation of bone density. J Biomech 21:825-837. 6. Heinrich C, Going S, Pamenter R, Perry S, Boyden T , Lohman T 1990 Bone mineral content of cyclically menstruating
14.
15.
female resistance and endurance trained athletes. Med Sci Sports Exerc 2 2 3 8 - 5 6 3 . Snow-Harter C , Bouxsein M, Lewis B, Charette S, Weinstein P, Marcus R 1990 Muscle strength as a predictor of bone mineral density in young women. J Bone Miner Res 5589595. Nieman D 1986 The Sports Medicine Fitness Course. Bull Publishing, Palo Alto, California. Heyward V 1984 Designs for Fitness. Burgess Publishing, Minneapolis, Minnesota. AAHPERD 1984 Technical Manual: Health Related Physical Fitness. American Alliance for Health, Physical Education, Recreation, and Dance, Reston, California. Dalsky G P , Stocke KS, Ehsani AA, Slatopolsky E, Lee WC, Birge SJ 1988 Weight-bearing exercise training and lumbar bone mineral content in postmenopausal women. Ann Intern Med 1082324-828. Gleeson PB, Protas E J , LeBlanc AD, Schneider VS, Evans H J 1990 Effects of weight lifting on bone mineral density in premenopausal women. J Bone Miner Res 5153-158. Rockwell J , Sorensen A, Baker S , Leahey D, Stock J , Michaels J, Baran D 1990 Weight training decreases vertebral bone density in premenopausal women: A prospective study. J Clin Endocrinol Metab 71:988-993. Jaworski Z 1976 Parameters and indices of bone resorption. In: Meunier P (ed) Bone Histomorphometry, 2nd International Workshop. Armour Montague, Paris, pp. 193-200. Cavanaugh DJ, Cann CE 1988 Brisk walking does not stop bone loss in postmenopausal women. Bone 9:201-204.
Address reprint requests to: Robert Marcus, M.D. GRECC, 182-B VA Medical Center Palo Alto, CA 94304 Received for publication June 19, 1991; in revised form January 22, 1992; accepted February 7, 1992.
