Review Article Osteoporosis - Springer Link

2 downloads 0 Views 132KB Size Report
replacement therapy (HRT) are the main recommended prophylactic and therapeutic regimens for osteoporosis in postmenopausal women [20]. An additional or.
Osteoporos Int (1999) 9:1–12 ß 1999 International Osteoporosis Foundation and National Osteoporosis Foundation

Osteoporosis International

Review Article The Effect of Exercise Training Programs on Bone Mass: A Metaanalysis of Published Controlled Trials in Pre- and Postmenopausal Women I. Wolff, J. J. van Croonenborg, H. C. G. Kemper, P. J. Kostense and J. W. R. Twisk Institute for Research in Extramural Medicine, Vrije Universiteit, Amsterdam, The Netherlands

Abstract. With the aging of the population, the medical and social costs of skeletal fragility leading to fractures will cause an immense burden on society unless effective prophylactic and therapeutic regimens can be developed. Exercise is suggested as a possible regimen against involutional bone loss. The purpose of the present metaanalysis is to address a quantitative review of the randomized controlled trials (RCTs) and nonrandomized controlled trials (CTs) on the effects of exercise training programs on bone mass, measured as bone mineral density (BMD) or bone mineral content (BMC), of the lumbar spine (LS) and the femoral neck (FN) in pre- and postmenopausal women. The literature from 1966 through December 1996 was searched for published RCTs and CTs. Study treatment effect is defined as the difference between percentage change in bone mass per year in the training group and the control group. Overall treatment effects (OTs) with the 95% confidence intervals of these study treatment effects were calculated using inverse-variance weighting. Of the 62 articles identified, 25 met the inclusion criteria and were maintained for further analyses. The weighted OTs for the RCTs showed very consistently that the exercise training programs prevented or reversed almost 1% of bone loss per year in both LS and FN for both pre- and postmenopausal women. The two OTs that could be calculated for strength training programs did not reach significance. The OTs for the CTs were almost twice as

Correspondence and offprint requests to: Prof. Dr H. C. G. Kemper, Institute for Research in Extramural Medicine (EMGO), Faculty of Medicine, Vrije Universiteit, Van der Boechorststraat 7, 1081 BT, Amsterdam, The Netherland. Tel: + 31 20 4448405. Fax: + 31 20 4448181. E-mail: [email protected].

high as those for the RCTs, which gives an indication of the confounding introduced by the nonrandom allocation of the subjects to groups. Keywords: Bone mass; Exercise; Meta-analysis; Osteoporosis; Women

Introduction The spate of literature on the topic of osteoporosis reflects the impact of this disease nowadays. Concern seems justified, because of the exponential increase in fracture incidence with age [1]. This incidence is increasing even more than would be expected from demographic changes in age, indicating a decrease in bone quality from generation to generation [2,3]. Therefore with the aging of the population, the medical, social and economic costs of skeletal fragility leading to fractures is certain to grow [1,4,5] unless effective prophylactic and therapeutic regimens can be developed. The chance that osteoporotic fractures will develop bears an important relation to bone mass [6–13], which is usually measured and expressed as bone mineral content (BMC) or bone mineral density (BMD). Although other factors besides low bone mass contribute to the risk of fracture, such as the geometric and microarchitectural properties of the bone [14,15] and, notably, fall-related factors [16], low-trauma fractures rarely occur in the absence of reduction in bone mass [1,10]. The bone mass present at a given time in life is determined by the factors that influence the gain of bone during growth and those that influence bone loss in later

2

life. These factors include genetics, hormonal status and lifestyle factors as nutrition and physical activity [4,17– 19]. Calcium and vitamin D supplementation and hormone replacement therapy (HRT) are the main recommended prophylactic and therapeutic regimens for osteoporosis in postmenopausal women [20]. An additional or alternative strategy exists in the shape of physical activity. Evidence demonstrating the importance of exercise for the skeleton comes from immobilization and bedrest studies [21,22], cross-sectional studies showing that physically active subjects have significant higher BMD than age-matched sedentary controls [for overview see, for example, 18,23–25], and within-subject studies indicating that athletes have greater bone mass in their dominant playing extremity than in their nondominant one [26,27]. The only study design that allows causal inferences, however, is the controlled intervention trial, especially when the groups are formed by randomization. Various clinical trials examining the effects of exercise training programs on bone mass have been reported. Narrative reviews mostly come to the conclusion that exercise training programs may maintain or improve the bone mass [24,25,28–31] or provide conflicting results [18,23,32]. A meta-analytic review gives the opportunity to combine the results of these studies statistically and to quantify the effects of exercise training programs on bone mass. The training and control groups can be formed either by randomization or by a form of non-random allocation. In exercise training studies allocation to the groups by choice is quite usual because of its potential positive effect on the compliance, which is often a problem, especially in training programs of longer duration. However, randomized controlled trials offer the best possibility for a valid evaluation of the treatment effect, while non-random allocation can introduce confounding by self-selection. Comparing the combined study treatment effects of the randomized controlled trials (RCTs) and non-randomized controlled trials (CTs) might reveal the effects of the confounding introduced by the non-random allocation. Therefore the purpose of the present study is to review – using meta-analytic techniques – the published RCTs and CTs on the effects of exercise training programs on bone mass (expressed as BMD or BMC) of pre- and postmenopausal women.

Methods Literature Search The goal of the literature search was to identify all available original articles reporting RCTs or CTs on the effects of exercise training programs on bone mass. The search for literature was limited to studies published in journals between 1966 and December 1996. Potential

I. Wolff et al.

studies were selected as a RCT if they met the criteria for a RCT according to the Cochrane Collaboration: (1) the study is on human subjects; (2) it is prospective in nature; (3) two or more interventions are compared with each other; (4) the assignment to a particular group is random. When the first three criteria were met, but the groups were formed by non-random allocation, then the study was selected as a CT. First an extensive computer search was performed using MEDLINE with the following keywords: Set A: ‘bone mass’ or ‘bone density’ or ‘bone mineral density’ or ‘bone mineral content’ Set B: ‘exercise’ or ‘exercise therapy’ or ‘training’ or ‘physical fitness’ or ‘physical activity’ Subsequently both sets were combined in the following way: search A and B. The yield of papers for set A was 10 113, for set B 125 701, and for set A and set B combined 925. The abstracts of the papers (or when the information was not available or not conclusive from the abstracts, the whole papers) were reviewed to see whether they met the mentioned selection criteria. On completion of the computer search, the reference lists were carefully cross-checked to identify additional eligible trials. In addition a handsearch was conducted on 1996 issues of the journals Calcified Tissue International and Journal of Bone and Mineral Research. Studies published as abstracts and unpublished data were not selected. Subsequently, studies were excluded from the analyses if they did not meet the following inclusion criteria: (1) The length of the training program was 16 weeks or more, because the formation of bone is a slow process taking at least several months to become complete. (2) Relevant outcome measure was BMD or BMC measured with dual-photon absorptiometry (DPA) or dual-energy X-ray absorptiometry (DXA); these methods are most conventional and measure precise locations of the bone, instead of a large region such as the whole trunk. Studies that used quantitative computed tomography (QCT) are not included, because QCT measures pure trabecular bone, while DPA and DXA measure both trabecular and cortical bone. (3) Location of measurement was the lumbar spine (LS) or femoral neck (FN); these two locations are most at risk for osteoporotic fractures, together with the radius. (4) Separate analyses were carried out for pre- and postmenopausal women, because of the rapid bone loss associated with estrogen deficiency after the menopause. (5) Separate analyses were carried out for men and women. (6) There was adequate data representation; studies were excluded that did not appropriately report sufficient information to allow computation of a study treatment effect and variance.

Data Abstraction The selected articles were reviewed on the inclusion criteria by one of the first two authors (I.W. and J.C.).

The Effect of Exercise Training Programs on Bone Mass

The inclusion or exclusion of the articles was subsequently discussed. Thereupon, for each included trial information was collected about the number of subjects in the training and control group; subject characteristics (mean age, baseline bone mass, calcium intake (when reported) and/or calcium supplementation; in postmenopausal women the use of hormone replacement therapy; training program characteristics (type of exercise: strength or endurance; length of the program; frequency, duration and intensity (when reported) of the training); measurement characteristics (location: lumbar spine or femoral neck; method: DPA or DXA; outcome: BMC or BMD). If both BMC and BMD were measured, only the BMD results were used in the meta-analyses. Also information required for the computation of the study treatment effect was abstracted, to wit, the changes between the pre- and postmenopausal values for both groups and/or the difference between those two changes. In addition information about the variance of these changes or of the difference between them was required. If necessary, i.e., if no standard deviations (SD) or standard errors (SE) were reported in the article, then, if possible, the SE was calculated from the confidence interval (CI), the 5th–95th percentile, the t or F value, or the p value. Because of their research design, some investigations enabled the calculation of multiple study treatment effects. If a study contained more than one experimental group, then each group mean (compared with the control group mean) was included and treated as a single datapoint. Secondly, if an article reported results for both LS and FN, then that article yielded study treatment effects for both locations.

3

The corresponding 95% confidence interval was constructed for the overall treatment effect by: 95% CI = OT ± 1.96 6 H(vOT) where vOT is the variance of the overall treatment effect, which is calculated using the following formula: vOT = 1/Swi Separate analyses were performed for the RCTs and CTs and, within this distinction, studies on pre- and postmenopausal women were also analyzed separately. If possible (minimum 3 studies), additional separate analyses were executed for strength and endurance training programs. All analyses were performed separately for the locations LS and FN. Results were considered significant at p50.05.

Results Sixty-two articles examining the effects of exercise training programs on bone mass were located [34–95]. Four of them examined only men [34–37]. Fifty-eight investigations examined the effects of training programs Table 1. List of inclusion criteria and studies excluded on the basis of these criteria Criteria

Excluded studies

1. Length of training program > 16 weeks 2. Relevant outcome measure BMD or BMC measured with DPA or DXA

RCTs: Beverly 1989 [38], Svendsen 1993 [39] RCTs: Chow 1987 [40], Revel 1993 [41] CTs: Ayalon 1987 [42], Cavanaugh 1988 [43], Chow 1987 [44], Dilsen 1989 [45], Jones 1991 [46] RCTs: Blumenthal 1991 [47], Preisinger 1995 [48], Sandler 1987 [49] CTs: Aloia 1978 [50], Heinonen 1996 [51], Prince 1991 [52], Rikli 1990 [53], Rundgren 1984 [54], Simkin 1987 [55], Smith 1981 [56], Smith 1989 [57], White 1984 [58] CTs: Krolner 1983 [59], Peterson 1991 [60] RCTs: Greendale 1993 [61]

Statistics From those studies that were included in the metaanalysis, study treatment effects (Ti) and the accompanying standard errors were calculated. Study treatment effect is defined as the difference between the percentage change per year in bone mass in the training group and in the control group. A positive figure indicates a protective effect of exercise. The expression of the study treatment effect as a percentage provides a standardization that allows the combining of study treatment effects of studies that differ in measurement outcome (BMD and BMC). The expression per year allows the combining of study treatment effects of studies that differ in the length of the training program. Overall treatment effects (OT) were calculated using inverse variance weighting, according to the following formula [33]: OT = S(wi 6 Ti)/Swi where wi is the weighting factor, which is the inverse of the variance of the study treatment effect (vTi): wi = 1/vTi

3. Location of measurement: lumbar spine or femoral neck

4. Separate analyses for preand postmenopausal women 5. Separate analyses for men and women 6. Adequate data representation

RCTs: Lohman 1995 [62], McCartney 1995 [63], Smidt 1992 [64] CTs: Hatard 1996 [65], Heikkinen 1991 [66], Nichols 1994 [67]

BMD, bone mineral density; BMC, bone mineral content; DPA, dual photon absorptiometry; DXA, dual-energy X-ray absorptiometry; RCT, randomized controlled trial; CT, non-randomized controlled trial.

4

I. Wolff et al.

Table 2. Selected study characteristics and the study treatment effects (Ti) with the accompanying weighting factors (wi) for the included randomized controlled trials First author and year of publication

Sample size (n)a

Mean age (yr)

Measurement methodc

Calcium intaked (mg/day)

HRTe

% change per yearf

Tig (%/yr)

wi h

T: 16/w high impact exercise + 50 heel drops/day at home C: 16/w low impact exercise + flexibility exercises at home

DXA

T: 1110 C: 1042



LS:

T:–3.42 C:–1.00 T: 0.11 C:–0.79

–2.42

0.111

0.90

2.159

Weightbearing activities 1h 36/w at 60–70% HRR + flexibility exercises

DXA

T: 0.55 C:–1.29 T: 0.27 C:–0.40

1.84

1.411

0.67

0.708

12

L: low impact activities 1h 36/w H: high impact activities 1h 36/w with peak forces 26body weight

DPA

L: 861 H: 935 C: 1039

T2 C1

LS:

L: 0.00 H: 1.71 C:–6.87

6.09 7.80

0.189 0.174

Length of training (months)

Training descriptionb

Postmenopausal women Endurance training Bassey T20 1995 [71] C24

Bravo 1996 [72]

T61 C63

54 55

59.6 59.9

12

12

FN:

T: 46 C: 48

T22 C16

LS: FN:

Grove 1992 [73]

L5 H5 C5

56.6 54.0 56.0

Hatori 1993 [74]

M9 H12 C9

58 56 58

7

M: 36/w 30min walking at 90% of the HR at anaerobic threshold H: 36/w 30min walking at 110% of the HR at anaerobic threshold

DXA

n.r



LS:

M:–1.71 H: 1.89 C:–2.91

1.2 4.8

0.204 0.251

Lau 1992 [75]

T15 C12

76 75

10

1006stepping up and down a block + 15 min exercise moving the upper trunk 46/w

DXA

T: 248+ 800S C: 275+ 800S

n.r

LS:

T:–1.32 C:–0.10 T: 6.0 C:-4.2

–1.22

1.209

10.2

0.637

See above

DXA

T: 259 C: 253

n.r

Lau 1992 [75]

T11 C12

79 75

10

FN:

LS: FN:

Martin 1993 [76]

Prince 1995 [77]

M20 L16 C19

60.3 57.8 56.7

12

T42? C42?

63 62

24

Strength training Kerr T23 1996a [78]

M: 30 min treadmill running at 70–85% HRmax 36/w L: 45 min treadmill running at 70–85% HRmax 36/w

DPA

Weightbearing exercise 2 h/w + walking 2 h/w at 60% HRpeak

DXA

M+L+C:

T:–2.28 C:–3.0 T:–7.92 C:-1.32

0.72 –6.6

0.843 0.392



LS:

M:–0.48 L: 0.81 C:–0.61

0.13 1.42

0.786 0.556

T: 919+ 1000S C: 822+ 1000S



LS:

T: 0.76 C:–0.10 T: 0.28 C:–0.18

0.86

7.987

0.46

6.427

1000S

FN:

55.7

12

Unilateral resistance training 3 sets 8 RM exercises of upper and lower limb 36/w 20–30min; other site served as control site

DXA

n.r



FN:

T: 0.0 C:–0.4

0.4

1.496

Kerr 1996b [78]

T19

58.4

12

Unilateral resistance training 3 sets 20 RM exercises of upper and lower limb 36/w 20–30min; other site served as control site

DXA

n.r



FN:

T: 0.2 C:–1.0

1.2

0.697

Nelson 1994 [79]

T20 C19

57.3 61.1

12

High-intensity strength training, various exercises: 3 sets of 8 reps. at 80% 1 RM 26/w 45 min

DXA

T: 931 C: 908



LS:

T: 1.0 C:–1.8 T: 0.9 C:–2.5

2.8

0.772

3.4

0.559

FN:

Notelovitz 1991 [80]

T9 C11

43.3 46.2

12

Resistance training 5 stations 8 RM 36/w 15–20 min

DPA

T: ±1400

T9 C11

LS:

T: 8.3 C: 1.5

6.8

0.051

Pruitt 1995 [81]

L7 H8 C11

67.6 67.0 69.6

12

10 resistance exercises 36/w 60–75 min L: 3 sets 14 reps. at 40% 1 RM H: 1 set 14 reps. at 40% 1 RM + 2 sets 7 reps. at 80%1RM

DXA

L: 701 H: 580 C: 660 +500S

L7 H4 C4

LS:

L: 0.5 H: 0.7 C:–0.1 L: 1.8 H:–0.2 C: 0.9

0.6 0.8

0.783 0.918

0.9 –1.1

0.290 0.492

Back strengthening exercises: lifting back pack 30% max isometric back muscles strength, in prone position, 56/w 106/day

DPA

T44+C43 0–800 T56+C57 >800



–0.2

4.052

Sinaki 1989 [82]

T34 C31

55.6 56.5

24

FN:

LS:

T:–1.4 C:–1.2

Continued over

The Effect of Exercise Training Programs on Bone Mass

5

Table 2. continued. First author and year of publication

Sample size (n)a

Mean age (yr)

Length of training (months)

Measurement methodc

Calcium intaked (mg/day)

HRTe

% change per yearf

Tig (%/yr)

wi h

T:1 h/w high-impact exercise + 50 jumps/day at home C: 1 h/w low impact exercise + arm exercises at home

DXA

T>1000 C>1000

n.a

LS:

T: 1.01 C: 1.5 T: 2.38 C:–1.83

–0.49

0.415

4.21

0.193

T: 1 h high-impact aerobics at 70–85% HRmax + weight training 36/w C: choice: no exercise or 30 min stretching 26/w

DXA

T: 0.65 C: 0.1 T: 0.25 C:–0.95

0.55

8.924

1.2

3.244

Progressive high-impact exercises: aerobics or steps 36/w 60 min

DXA

T: C: T: C:

1.0

7.246

0.67

7.091

3.0 3.15

0.687 0.730

Training descriptionb

Premenopausal women Bassey 1994 [83]

Friedlander 1995 [84]

Heinonen 1996 [85]

T14 C13

T32 C31

T39 C45

32.0 29.8

28.0 30.1

39 39

6

24

18

FN:

T16+C16 ±1500 T16+C15: 300–1000

n.a

T: 1125 C: 1102

n.a

LS: FN:

LS: FN:

Snow-Harter 1992 [86]

W12 R10 C8

19.9

8

W: weight training: 14 exercises 3 sets of 8–12 reps. at 85%1RM 36/w R: running 70–80% HRmax 36/w, mileage increased

DXA

W+R +C: 500S

n.a

LS:

1.47 0.47 1.07 0.4

W: 1.8 R: 1.95 C:–1.2

n.r.; not reported; n.a., not applicable. a T, training group; C, control group. For example, T20 indicates 20 subjects in the training group. b 6/w, times per week; h/w, hour per week; HRR, heart rate reserve; HRmax, maximal heart rate; RM, repetition maximum; reps., repetitions. c DPA, dual photon absorptiometry; DXA, dual-energy X-ray absorptiometry. d S, supplement. e HRT, hormone replacement therapy; –, no subjects used HRT. f LS, lumbar spine; FN, femoral neck. g Ti, study treatment effect calculated as the % change per year of the training group minus % change per year of the control group. h wi, weighting factor calculated as 1/vTi.

in women. Twenty-eight of these articles contained RCTs and 30 contained CTs. Following the application of the inclusion criteria, 16 articles with RCTs [71–86] and 9 articles with CTs [87–95] were maintained for further analyses. The studies excluded are mentioned in Table 1, listed under the first criterion that they did not meet [38–67]. Blumenthal et al. [68], Seidl et al. [69] and Smith et al. [70] were excluded because they reported on the same data set as Blumenthal et al. [47], Hatard et al. [65] and Smith et al. [57], respectively. The 25 investigations included together yielded 53 study treatment effects: 34 for the RCTs and 19 for the CTs. Tables 2 and 3 briefly summarize these RCTs and CTs respectively, and indicate the study treatment effects and weighting factors calculated. The length of the training programs ranged from 6 to 24 months. Six studies allowed or included HRT and 16 studies allowed or included calcium supplementation. Forty-two study treatment effects were positive, indicating less reduction or more gain in bone mass for the training group compared with the control group. Eleven study treatment effects were negative, indicating more reduction or less gain in bone mass for the training group compared with the control group. Only 19 of the 53 study treatment effects reached significance, of which one was significantly negative.

The overall treatment effects for RCTs are found in Table 4 and for CTs in Table 5. With the exception of differences found for strength training in postmenopausal women in both LS and FN for the RCTs and LS in the premenopausal women for the CTs, significant differences were noted across all categories. For the RCTs (Table 4) significant OTs of 0.84%/year and 0.89%/year were observed for the LS and FN bone locations. The partition of these OTs for pre- and postmenopausal women shows that training caused significant OTs in both premenopausal (LS: 0.91%/ year and FN: 0.90%/year) and postmenopausal women (LS: 0.79%/year and FN: 0.89%/year). The separate analyses for training type in postmenopausal women showed that the OTs of the endurance training programs were significant in both LS (0.96%/year) and FN (0.90%/year), but that strength training did not result in significant OTs. For the CTs also (Table 5) significant OTs were observed (LS: 1.85%/year and FN: 1.39%/year). In contrast to the RCTs, the CTs showed significant results only in the postmenopausal women for both LS (2.40%/ year) and FN (1.68%/year) and not for the three studies in premenopausal women. When the only strength training study was subtracted, significant OTs for endurance training were found in postmenopausal women (LS: 2.25%/year and FN: 1.86%/year).

6

I. Wolff et al.

Table 3. Selected study characteristics and study treatment effects (Ti) with the accompanying weighting factors (wi) for the included nonrandomized controlled trials First author and year of publication

Sample size (n)a

Mean age (yr)

Length of training (months)

Training descriptionb

Measurement methodc

Calcium intaked (mg/day)

HRTe

% change per yearf

DPA

T: 1500 C: 1433



LS:

Tig (%/yr)

wi h

Postmenopausal women Endurance training Bloomfield T7 1993 [87] C7

62.0 58.8

8

Cycle ergometry, 36/w ±30 min at 60–80% HRmax

FN:

Caplan 1993 [88]

T19 C11

66.4 65.4

24

T: 970 C: 1030



LS:

T: 5.33 C:–3.66 T: 3.77 C:–1.11

8.99

0.165

4.88

0.090

T:–0.4 C:–1.9 T: 1.45

1.5

2.372

0.25

0.703

Aerobic weight-bearing exercise, 26/w 60 min, instructed to work out at least 16/w extra

DPA

Walking and jogging at 60–70% VO2max 36/w 50–60 min After 3 months also stair climbing 8 reps. of 88 steps

DPA

T:±1500 C:±1500

T4 C3

LS:

T: 6.93 C:–1.87

8.80

0.176

2 months flexibility exercises then 9 months walking, jogging and stair climbing 3–56/w ±45 min

DXA

T:1390 C:1308



LS:

T: 2.66 C: 0.24 T: 3.58 C:–0.90

2.42

1.359

4.48

1.004

See above

DXA

2.32

0.317

–0.08

0.730

–2.44

0.145

–0.20

0.254

2.41 0.44 2.96 0.87

1.97

0.130

2.09

0.664

T: 2.13 C:–4.8 T:–3.6 C:–1.07

6.93

0.155

–2.53

0.141

FN:

C: 1.2 Dalsky 1988 [89]

T17 C18

61.6 62.2

Kohrt 1995 [90]

T8 C8

65 66

Kohrt 1995 [90]

T8 C8

66 67

9

11

11

FN:

T: 1478 C: 1574

T8 C8

LS: FN:

Nelson 1991 [91]

Nelson 1991 [91]

T9 C9

T9 C9

60.2

60.2

12

12

Supervised walking 46/w 50 min, after 4 w wearing loaded belts (3.1 kg), 75–80% HRmax

DPA

See above

DPA

T+C: 761



LS: FN:

T+C: 1462



LS: FN:

Strength training Pruitt T13 1992 [92] C9 T16 C9

53.6 55.6

9

Weight-lifting exercises of the arms, legs and trunk 36/w 60 min 10 RM

DPA

T: 1008 C: 812



LS: FN:

T: C: T: C:

8.92 6.60 2.25 2.33

T:–0.85 C: 1.60 T:–1.19 C:–1.00 T: C: T: C:

Premenopausal women Gleeson 1990 [93]

T34 C34

33.4 32.7

12

Rockwell 1990 [94]

T10 C7

36.2 40.4

9

Vuori 1994 [95]

T12

22.0

12

Nautilus weight-lifting program DPA upper and lower extremities, 36/w 2 sets of 20 reps. at 60% 1 RM 8-station resistance-training DXA circuit 26/w ±45 min 1 set of 12 reps. at ±70% 1 RM at each station

T+C: 500S

n.a

LS:

T: 0.82 C:–0.52

1.34

1.442

T+C: 500S

n.a

LS:

T:–5.28 C:–1.12

–4.16

0.071

Unilateral strength training using legpress 56/w 5 sets of 10 reps. at 80% 1 RM

n.r

n.a

LS:

T: C: T: C:

0.6

1.218

0.1

0.813

DXA

FN:

2.0 1.4 1.1 1.0

n.r., not reported; n.a., not applicable. a T, training group; C, control group; For example, T20 indicates 20 subjects in the training group. b 6/w, times per week; h/w, hours per week; HRR, heart rate reserve; HRmax, maximal heart rate; VO2max, maximal oxygen uptake; RM, repetition maximum; reps. repetitions. c DPA, dual photon absorptiometry; DXA, dual energy X-ray absorptiometry. d S, supplement. e HRT, hormone replacement therapy; –, no subjects used HRT. f LS, lumbar spine; FN, femoral neck. g Ti: study treatment effect calculated as the % change per year of the training group minus % change per year of the control group. h wi, weighting factor calculated as 1/vTi.

The Effect of Exercise Training Programs on Bone Mass

7

Table 4. Overall treatment effects (OT) calculated with inverse variance weighting and 95% confidence intervals (95% CI) for the randomized controlled trials Lumbar spine

Femoral neck

Traininga

OT (%/yr)

Premenopausal women e+s

0.91*

0.44–1.37

Postmenopausal women e 0.96* s 0.44 e+s 0.79* Total

0.84*

95% CI

nb

Training

OT (%/yr)

95% CI

n

5

e+s

0.90*

0.29–1.50

3

0.43–1.49 –0.32–1.21 0.35–1.22

11 5 16

e s e+s

0.90* 0.86 0.89*

0.29–1.51 –0.18–1.91 0.36–1.42

5 5 10

0.53–1.16

21

Total

0.89*

0.50–1.29

13

a

e, endurance training; s, strength training. n, number of study treatment effects included in the analysis. *p50.05. b

Table 5. Overall treatment effects (OT) calculated with inverse variance weighting and 95% confidence intervals (95% CI) for the nonrandomized controlled trials Lumbar spine

Femoral neck

Traininga

OT (%/yr)

95%CI

nb

Premenopausal women e+s

0.90

–0.29–2.09

3

e+s

1.83–2.67

e s e+s

1.86*

0.80–2.91

1.68*

0.65–2.72

6 1 7

Total

1.39*

0.46–2.33

8

Postmenopausal women e 2.25* s e+s 2.40*

2.00–2.81

7 1 8

Total

1.59–2.11

11

1.85*

Training

OT (%/yr)

95% CI

n

1

a

e, endurance training; s: strength training. n, number of study treatment effects included in the analysis. *p50.05 b

Discussion The purpose of this meta-analysis was to quantify objectively the effects of exercise training programs on the bone mass of pre- and postmenopausal women. This is performed for the relevant bone locations lumbar spine and femoral neck. Studies with two types of research designs are included in the analyses: randomized controlled trials and non-randomized controlled trials. All the overall treatment effects for the RCTs are found to be around 0.9% per year (Table 4), meaning that exercise training programs prevented or reversed bone loss of almost 1% per year compared with the controls. The overall treatment effects were consistent for the lumbar spine and the femoral neck, and also for the pre- and postmenopausal women. The overall treatment effects for the CTs were mostly twice as high as those for the RCTs (Table 5). Of all study designs the RCT has the highest quality and controls most for confounding. The higher overall treatment effects for the CTs compared with RCTs indicate a potential overestimation of the true treatment

effect in the CTs as a consequence of the introduction of confounding by non-random allocation of the subjects to the groups. This indicates that results of research employing a non-randomized design must be interpreted cautiously. Treatment effects reported in these investigations could at least partly be a function of other variables (e.g., genetics, self-selection, previous experiences). Therefore the overall treatment effects of the RCTs are considered the best estimations of the true treatment effects. To generate the overall treatment effects inverse variance weighting was used [33]. This weighting method results in large contributions of studies with large sample sizes and small standard errors, and thus with high precision. A high precision is an indicator of the quality of the study. A weighting method with quality rating scores based upon criteria for methodologic assessment of the studies, which is also used often [e.g., 96], might assess just the quality (completeness) of documentation rather than the quality of the study itself. The range in study treatment effects found in this meta-analysis is very large, from –6.6% per year [75] to

8

10.2% per year [75], but the overall treatment effects are surprisingly consistent. This is a consequence of the inverse variance weighting, resulting in small weighting factors for the outlying values and the larger weighting factors for the modest study treatment effects. The most important narrative reviews on exercise and bone mass concluded that intervention studies show a positive effect of exercise on bone mass [24,25,29–31] or that intervention studies provide conflicting results [18,23,32]. All agree that the cross-sectional studies show larger positive effects. Gutin and Kasper [28] concluded that in general it seems bone mass can be enhanced by both strenuous aerobic exercise and strength training. It seems that mild general exercise such as walking is not effective in preventing postmenopausal bone loss or enhancing bone mass in younger age periods. The way in which exercise is thought to act on the skeleton is through gravitational forces or muscle pull producing strains within the skeleton which are perceived by bone cells as osteogenic. If a strain is detected as greater than the optimum strain, then bone formation will occur [97]. Animal studies have shown that unusual strain distributions, high strains and high strain rates seem to be particularly osteogenic [98–100]. The osteogenic response that follows exposure to such strains appears to saturate after only a few loading cycles [101]. Application to a human model implies that strength training programs with a large diversity in exercises instead of endurance training programs such as running should result in the greatest increase in skeletal density. However, an endurance training program involving aerobic exercises might also impose the necessary strains, strain rates and unusual strain distributions. No large differences between the overall treatment effects for the endurance and strength training programs (only two OTs could be calculated for strength training) are found. The overall treatment effects for strength training programs did not reach significance. However, this can be a consequence of the small number of studies with strength training programs. Next to this, the strength training programs might not all have imposed the required unusual strain distributions, high strains and high strain rates. For example, the program of Sinaki et al. [82], which involved lifting a back pack at 30% maximal isometric strength in the prone position, 10 times a day for 5 days a week, might not have been diverse and intense enough. This study did indeed result in a treatment effect of –0.2% per year. Training programs that fulfil all the recommended requirements of high strain and strain rates may not be the most attractive ones for the elderly population, for the frail elderly they may even involve risks. It is very important to make the exercise programs as attractive as possible to assure high compliance and attendance, which are large problems in long-term exercise programs. The osteogenic effects of exercise training seems to be site-specific to the anatomic sites at which the mechanical strains occur [26,98,102]; thus it is essential

I. Wolff et al.

to measure the bone mass at the site of loading. Therefore no overall treatment effects are calculated for the radius, although this is a bone site at risk for osteoporotic fractures, because only a few studies specifically loaded the radius [42,51,55]. In most studies the training of the forearms is only a small part of the training program, if at all, and is hardly reported [47– 50,52,53,56–58]. The nature of meta-analysis implies that the metaanalysis itself inherits those limitations that exist in the literature. The included studies can differ greatly, resulting in heterogeneity, which is a common problem in meta-analyses. The large range in study treatment effects (as mentioned above) can be the result of heterogeneity between the studies. Important sources of heterogeneity in this meta-analysis will be discussed in the following section. There are large differences in type, duration, frequency and intensity of the training among the different training programs. Comparing the different programs with each other is hardly possible, especially with respect to the intensity, not the least because of the insufficient and non-standardized descriptions. To reduce the effect of this source of heterogeneity, separate analyses were performed for strength end endurance training programs. Other important sources of heterogeneity in this metaanalysis are the use of calcium and vitamin D supplementation and HRT. Both are established to have a positive influence on the bone mass and are often prescribed for prevention and therapy of osteoporosis [20]. Quite a few studies allowed or included these co-interventions as part of the program. Calcium supplementation was allowed or included, for all or a portion of the subjects, in 16 of the 25 studies [71, 72,75–77,80,81,83,84,86–88,90,91,93,94]. HRT was allowed in 6 of the 18 studies on postmenopausal women [69,70,77,78,88,90]. Excluding all these studies would result in too few datapoints to be able to perform a meta-analysis. Differences in the bone mass at baseline can also explain a part of the range in study treatment effects. It can be expected from other physiologic parameters that women with low bone mass at the start of the training program will show a greater increase in bone mass than those with a better bone status. One study [72] included only osteopenic women. The treatment effects of this study however were not deviating from the other studies. Also the age of the subjects can be important, because of the involutional bone loss that occurs with aging. The mean age of the postmenopausal subject groups is close to 55–60 years; only in the study by Lau et al. [75] was the mean age considerably higher (± 76 years). This study was performed with Chinese women with an extremely low dietary calcium intake (less than 400 mg/ day). One group received calcium supplements and loadbearing exercise, compared with a control group receiving only calcium supplements. The study treatment effect for the femoral neck was 10.2% per year, while for the groups without calcium supplementation

The Effect of Exercise Training Programs on Bone Mass

the study treatment effect was –6.6% per year. Although no such extreme values were found for the lumbar spine, this might indicate that the positive effect of exercise can only be established if sufficient calcium is available. Specker [103] concluded that a positive effect of physical activity appears to exist only at calcium intakes greater than 1000 mg/day. Body weight is positively related to bone mass [8,104–106], and therefore weight loss can be a confounder for the exercise groups in long-term interventions of this type. Weight loss due to the training program might cause an underestimation of the effect. Most studies reported only the baseline values for the body weight of the training and control groups and not the post-exercise value or the changes in body weight. The studies reporting the change in body weight did not indicate considerable changes in body weight in the training groups compared with the control groups [68,70,73,84,89,90,91,93]. Low compliance is a problem during exercise programs, especially in training programs of long duration. All studies but one analyzed only the subjects who were compliant (per treatment analysis), instead of using an intention-to-treat analysis strategy. This can result in an overestimation of the treatment effect. Twenty-one of the 25 studies mentioned the compliance. With a few exceptions the compliance was surprisingly high, ranging from 50% to 100%; while two-thirds of the studies reported a compliance 5 75%. It needs to be mentioned that expressing the OTs as percentage change per year is only correct if the kinetics of bone density changes are a linear function of time, which they probably are not. It might be possible that the bone accretion decreases when a given level is achieved and that further responses will be slower and of a smaller magnitude. This means that expressing the study treatment effects of short studies as percentage change per year can lead to an overestimation. Therefore the analyses were also performed without the studies that were shorter than 10 months [74,83,86,87,89,92,94]. This resulted in only small decreases in OTs for the RCTs and both small increases and decreases in OTs for the CTs. The maximal length of an exercise training program included in this meta-analysis was 2 years; therefore no inferences can be drawn regarding programs of a longer duration. It may well be that the OT observed after 6–24 months is in fact the maximal amount of bone loss that exercise can prevent after 1, 2 or 10 years. There is a need for studies with a duration of more than 5 years in postmenopausal women to evaluate the feasibility and relevance of the bone preservation by exercise. More than half of the study treatment effects (34 of the 53) included in this meta-analysis were not significant. This may partly be a consequence of the small sample sizes in the clinical trials, which result in a lack of power of these studies. The power of a study, which is the probability that a study of a given size would detect as statistically significant a real difference of a given magnitude, is commonly required to be

9

between 80% and 90% [107]. With small sample sizes there is a substantial chance that a statistically significant difference between the training and the control group is not found, while in fact there is such a difference (false negative result). The strength of a meta-analysis is that by combining the various studies, and thereby adding up the numbers of subjects, the power of the analysis increases. In this way a small but existing difference can be demonstrated. With realistic assumptions of the true effect of training programs on the bone mass and the accompanying standard deviation, computations can be made of necessary sample sizes. These assumptions can be based on the results of this meta-analysis. With a mean weighted difference (between the training and control groups) of about 0.9% per year and a standard deviation of about 4.5 this means for the RCTs that, for a power of between 80% and 90%, the sample size needs to be 400–500 subjects for each group [107]. None of the trials included fulfilled this high requirement of such large sample sizes. Some small studies included here may have yielded statistically significant results only because, by chance, the observed difference in the sample is much larger than the real difference (see, e.g., [73,87,92]. The importance of the finding of this meta-analysis, that bone mass was preserved in the exercising women, lies in the fact that bone density is closely related to probability of future fracture [6–13]. While an overall treatment effect of 0.9% per year may appear small, this could have a significant impact on the development of osteoporotic fractures in large populations. After the rapid bone loss during the first postmenopausal years, the rate of resorption returns to the rate found before the menopause, about 1% per year [18,108–110]. According to this meta-analysis, exercise training has the potential to counteract this loss completely. However, exercise will not have the power to prevent the accelerated bone loss seen in the first years after menopause. The effects of exercise training can be compared with the effects of other possible options for treatment and prevention of osteoporosis. Welten et al. [111] found a slightly higher overall effect of 1.3% per year for ~1000 mg/day calcium supplementation in premenopausal women. Cumming [112] found for postmenopausal women an overall effect of 0.8% per year for calcium supplementation. HRT is considered to prevent the rapid bone loss accompanying estrogen deficiency after menopause. However, side effects such as the menstrual bleeding, the necessity of long-term use and possibly a small increase in the risk of breast cancer result in many women not being compliant or not choosing to start using HRT. The benefits, side effects and risks of this therapy are extensively reviewed elsewhere [e.g., 113]. Next to the potential of exercise training to prevent or reverse the loss of bone by almost 1% per year, also risk factors for falls, such as low muscle mass and muscle strength, and poor balance and coordination [114], can be modified by exercise [79]. Falls are next to low bone mass, an important determinant for osteoporotic fractures [16]. Therefore, in contrast to pharmacologic and

10

nutritional approaches, exercise training has the potential to prevent osteoporotic fractures by simultaneously influencing multiple risk factors [115]. In conclusion, the important finding of this metaanalysis is that exercise training programs have the capability to prevent or reverse bone loss in the lumbar spine and the femoral neck of pre- and postmenopausal women. The overall treatment effects for studies with the highest quality, the RCTs, showed a modest but clinically relevant and consistent effect of 0.9% per year. The CTs resulted in overall treatment effects that were about twice as high, which gives an indication of the confounding introduced by the nonrandom allocation of the subjects. This meta-analysis shows that exercise training programs are worth considering in the prevention or treatment of osteoporosis.

References 1. Riggs BL, Melton LJ. The worldwide problem of osteoporosis: insights afforded by epidemiology. Bone 1995; 5(Suppl):S505– 11. 2. Kannus P, Parkkari J, Niemi S. Age-adjusted incidence of hip fractures. Lancet 1995;346:50–1. 3. Obrant KJ, Bengne´r U, Johnell O, Nilsson BE, Sernbo I. Increasing age-adjusted risk fragility fractures: a sign of increasing osteoporosis in successive generations? Calcif Tissue Int 1989;44:157–67. 4. World Health Organisation. Assessment of fracture risk and its application to screening for postmenopausal osteoporosis; report of a WHO study group. WHO Technical report series 843. Geneva: WHO, 1994. 5. Randell A, Sambrook PN, Nguyen TV, et al. Direct clinical and welfare costs of osteoporotic fractures in elderly men and women. Osteoporos Int 1995;5:427–32. 6. Cummings SR, Black DM, Nevitt MC, et al. Appendicular bone density and age predict hip fracture in women. JAMA 1990;263:665–8. 7. Cummings SR, Black DM, Nevitt MC, et al. Bone density at various sites for prediction of hip fractures. Lancet 1993;341:72–5. 8. Hui SL, Slemenda CW, Johnston CC. Age and bone mass as predictors of fracture in a prospective study. J Clin Invest 1988;81:1804–9. 9. Marshall D, Johnell O, Wedel H. Meta-analysis of how well measures of bone mineral density predict occurrence of osteoporotic fractures. BMJ 1996; 312:1254–9. 10. Melton LJ, Wahner HW, Richelson LS, O’Fallon WM, Riggs BL. Osteoporosis and the risk of hip fracture. Am J Epidemiol 1986;124:254–61. 11. Melton LJ, Atkinson AJ, O’Fallon WM, Wahner HW, Riggs BL. Longterm fracture prediction by bone mineral assessed at different skeletal sites. J Bone Miner Res 1993;8:1227–33. 12. Ross PD, Davis JW, Vogel JM, Wasnich RD. A critical review of bone mass and the risk of fractures in osteoporosis. Calcif Tissue Int 1990;46:149–61. 13. Wasnich R. Bone mass measurement: prediction of risk. Am J Med 1993;95:S6–10. 14. Faulkner KG, Cummings SR, Black D, Palermo L, Glue¨r C-C, Genant HK. Simple measurement of femoral geometry predicts hip fracture. J Bone Miner Res 1993;8:1211–7. 15. Parfitt AM. Implications of architecture for the pathogenesis and prevention of vertebral fracture. Bone 1992;13:S41–7. 16. Dargeant-Molina P, Favier F, Grandjean H et al. Fall-related factors and risk of hip fracture: the EPIDOS prospective study. Lancet 1996;348:145–9. 17. Krall EA, Dawson-Hughes B. Heritable and life-style determinants of bone mineral density. J Bone Miner Res 1993;8:1–9.

I. Wolff et al. 18. Snow CM, Shaw JM, Matkin CC. Physical activity and risk for osteoporosis. In: Marcus R, Feldman D, Kelsey J, editors. Osteoporosis. San Diego: Academic Press, 1996:511–28. 19. Salamone LM, Glynn NW, Black DM et al. Determinants of premenopausal bone mineral density: the interplay of genetic and lifestyle factors. J Bone Miner Res 1996;11:1557–65. 20. Lane JM, Riley EH, Wirganowicz PZ. Osteoporosis: diagnosis and treatment. J Bone Joint Surg Am 1996;7:618–32. 21. Krolner B, Toft B. Vertebral bone loss: an unheeded side effect of therapeutic bed rest. Clin Sci 1983;64:537–40. 22. Whedon GD. Disuse osteoporosis: physiological aspects. Calcif Tissue Int 1984;36:S146–50. 23. Bouxsein ML, Marcus R. Overview of exercise and bone mass. Rheum Dis Clin North Am 1994;20:787–801. 24. Chilibeck PD, Sale DG, Webber CE. Exercise and bone mineral density. Sports Med 1995;19:103–22. 25. Forwood MR, Burr DB. Physical activity and bone mass: exercises in futility? Bone Miner 1993;21:89–112. 26. Haapasalo H, Kannus P, Sieva¨nen A, Oja P, Vuori I. Long-term unilateral loading and bone mineral density and content in female squash players. Calcif Tissue Int 1994;54:249–55. 27. Huddleston AL, Rockwell D, Kulund DN, Harrison B. Bone mass in lifetime tennis athletes. JAMA 1980;244:1107–9. 28. Gutin B, Kasper MJ. Can vigorous exercise play a role in osteoporosis prevention? A review. Osteoporos Int 1992;2:55– 69. 29. Mosekilde L. Osteoporosis and exercise. Bone 1995;17:193–5. 30. Sinaki M. Exercise and osteoporosis. Arch Phys Med Rehabil 1989;70:220–9. 31. Smith EL, Gilligan C. Physical activity effects on bone metabolism. Calcif Tissue Int 1991;49(Supp):S50–4. 32. Marcus R, Drinkwater B, Dalsky G, et al. Osteoporosis and exercise in women. Med Sci Sports Exerc 1992;24:S301–7. 33. Shadish WR, Haddock CK. Combining estimates of effect size. In: Cooper H, Hedges LV, editors. The handbook of research synthesis. New York: Russell Sage Foundation, 1994:261–81. 34. Cohen B, Milett PJ, Mist B, Laskey MA, Rushton N. Effect of exercise training programme on bone mineral density in novice college rowers. Br J Sports Med 1995;29:85–8. 35. Menkes A, Mazel S, Redmond RA et al. Strength training increases regional bone mineral density and bone remodeling in middle-aged and older men. J Appl Physiol 1993; 74(5):2478-2484. 36. Ryan AS, Treuth MS, Rubin MA, et al. Effects of strength training on bone mineral density: hormonal and bone turnover relationships. J Appl Physiol 1994;77:1678–84. 37. Williams JA, Wagner J, Wasnich R, Heilbrun L. The effect of long-distance running upon appendicular bone mineral content. Med Sci Sports Exerc 1984;16:223–7. 38. Berverly MC, Rider TA, Evans MJ, Smith R. Local bone mineral response to brief exercise that stresses the skeleton. BMJ 1989;299:233–5. 39. Svendsen O-L, Hassanger C, Christiansen C. Effect of an energy-restrictive diet, with or without exercise, on lean tissue mass, resting metabolic rate, cardiovascular risk factors, and bone in overweight postmenopausal women. Am J Med 1993;95:131–40. 40. Chow RK, Harrison JE, Notarius C. Effect of two randomised exercise programmes on bone mass of healthy postmenopausal women. BMJ 1987;295:1441–4. 41. Revel M, Mayoux-Benhamou MA, Rabourdin JP, Bagheri F, Roux C. One-year psoas training can prevent lumbar bone loss in postmenopausal women: a randomized controlled trial. Calcif Tissue Int 1993;53:307–11. 42. Ayalon J, Simkin A, Leichter I, Raifmann S. Dynamic bone loading exercises for postmenopausal women: effect on the density of the distal radius. Arch Phys Med Rehabil 1987;68:280–3. 43. Cavanaugh DJ, Cann CE. Brisk walking does not stop bone loss in postmenopausal women. Bone 1988;9:201–4. 44. Chow RK, Harrison JE, Sturtridge W, et al. The effect of exercise on bone mass of osteoporotic patients on fluoride treatment. Clin Invest Med 1987;10:59–63.

The Effect of Exercise Training Programs on Bone Mass 45. Dilsen G, Berker C, Oral A, Varan G. The role of physical exercise in prevention and management of osteoporosis. Clin Rheum 1989;8(Suppl 2):70–5. 46. Jones PRM, Hardman AE, Hudson A, Norgan NG. Influence of brisk walking on the broadband ultrasonic attenuation of the calcaneus in previously sedentary women aged 30–61 years. Calcif Tissue Int 1991;49:112–15. 47. Blumenthal JA, Emery CF, Madden DJ, et al. Effects of exercise training on bone density in older men and women. J Am Geriatr Soc 1991;39:1065–70. 48. Preisinger E, Alacamlioglu Y, Pils K, Saradeth T, Schneider B. Therapeutic exercise in the prevention of bone loss: a controlled trial with women after menopause. Am J Phys Rehabil 1995;74:120–3. 49. Sandler RB, Cauley JA, Hom DL, Sashin D, Kriska AM. The effects of walking on the cross-sectional dimensions of the radius in postmenopausal women. Calcif Tissue Int 1987;41:65– 9. 50. Aloia JF, Cohn SH, Ostuni JA, Cane R, Ellis K. Prevention of involutional bone loss by exercise. Intern Med 1978;89:356–8. 51. Heinonen A, Sieva¨nen H, Kannus P, Oja P, Vuori I. Effects of unilateral strength training and detraining on bone mineral mass and estimated mechanical characteristics of the upper limb bones in young women. J Bone Miner Res 1996;11:490–501. 52. Prince R, Smith M, Dick IM, et al. Prevention of postmenopausal osteoporosis: a comparative study of exercise, calcium supplementation, and hormone-replacement therapy. N Engl J Med 1991;325:1189–95. 53. Rikli RE, McManis BG. Effects of exercise on bone mineral content in postmenopausal women. Res Q Exerc Sports 1990;61:243–9. 54. Rundgren A, Aniansson A, Ljungberg P, Wetterqvist H. Effects of a training programme for elderly people on mineral content of the heel bone. Arch Gerontol Geriatr 1984;3:243–8. 55. Simkin A, Ayalon J, Leichter I. Increased trabecular bone density due to bone-loading exercises in postmenopausal osteoporotic women. Calcif Tissue Int 1987;40:59–63. 56. Smith EL, Reddan W, Smith PE. Physical activity and calcium modalities for bone mineral increase in aged women. Med Sci Sports Exerc 1981;13:60–4. 57. Smith EL, Gilligan C, McAdam M, Ensign CP, Smith PE. Deterring bone loss by exercise intervention in premenopausal and postmenopausal women. Calcif Tissue Int 1989;44:312–21. 58. White MK, Martin RB, Yeater RA, Butcher RL, Radin EL. The effects of exercise on the bones of postmenopausal women. Int Orthop 1984;7:209–14. 59. Krolner B, Toft B, Nielsen SP, Tondevold E. Physical exercise as prophylaxis against involutional vertebral bone loss: a controlled trial. Clin Sci 1983;64:541–6. 60. Peterson SE, Peterson MD, Raymond G, Gilligan C, Checovich MM, Smith EL. Muscular strength and bone density with weight training in middle-aged women. Med Sci Sports Exerc 1991;23:499–504. 61. Greendale GA, Hirsch SH, Hahn TJ. The effect of a weighted vest on perceived health status and bone density in older persons. Qual Life Res 1993;2:141-52. 62. Lohman T, Going S, Pamenter R, et al. Effects of resistance training on regional and total bone mineral density in premenopausal women: a randomized prospective study J Bone Miner Res 1995;10:1015–24. 63. McCartney N, Hicks AL, Martin J, Webber CE. Long-term resistance training in the elderly: effects on dynamic strength, exercise capacity, muscle and bone. J Gerontol 1995; 50A:B97– 104. 64. Smidt GL, Lin S-Y, O’Dwyer KD, Blanpied PR. The effect of high-intensity trunk exercise on bone mineral density of postmenopausal women. Spine 1992;17:280–5. 65. Hartard M, Haber P, Ilieva D, Preisinger E, Seidl G, Huber J. Systematic strength training as a model of therapeutic intervention: a controlled trial in postmenopausal women with osteopenia. Am J Phys Med Rehabil 1996;75:21–8. 66. Heikkinen J, Kurttila-Matero E, Kyllo¨nen E, Vuori J, Takala T,

11

67. 68. 69. 70. 71.

72. 73. 74. 75. 76. 77.

78. 79.

80. 81.

82.

83. 84.

85. 86.

87.

Va¨a¨na¨nen HK. Moderate exercise does not enhance the positive effect of estrogen on bone mineral density in postmenopausal women. Calcif Tissue Int 1991;49(Suppl):S83–4. Nichols DL, Sanborn CF, Bonnick SL, Ben-Ezra V, Gench B, DiMarco N. The effects of gymnastics training on bone mineral density. Med Sci Sports Exerc 1994;26:1220–25. Blumenthal JA, Emery CF, Madden DJ et al. Cardiovascular and behavioral effects of aerobic exercise training in healthy older men and women. J Gerontol 1989;44:M147–57. Seidl G, Kainberger F, Haber P, et al. Systematische Krafttraining in der Postmenopause: Begleitende densitometrische Kontrole mit DXA. Radiologe 1993;33:452–6. Smith EL, Smith PE, Ensign CJ, Shea MM. Bone involution decrease in exercising middle-aged women. Calcif Tissue Int 1984;36:S129–38. Bassey EJ, Ramsdale SJ. Weight-bearing exercise and ground reaction forces: a 12-month randomized controlled trial of effects on bone mineral density in healthy postmenopausal women. Bone 1995;16:469–76. Bravo G, Gauthier P, Roy P-M, et al. Impact of a 12-month exercise program on the physical and psychological health of osteopenic women. J Am Geriatr Soc 1996;44:756–62. Grove KA, Londeree BR. Bone density in postmenopausal women: high impact vs low impact exercise. Med Sci Sports Exerc 1992;24:1190–4. Hatori M, Hasegawa A, Adachi H, et al. The effects of walking at the anaerobic threshold level on vertebral bone loss in postmenopausal women. Calcif Tissue Int 1993;52:411–4. Lau EMC, Woo J, Leung PC, Swaminthan R, Leung D. The effects of calcium supplementation and exercise on bone density in elderly chinese women. Osteoporos Int 1992;2:168–73. Martin D, Notelovitz M. Effects of aerobic training on bone mineral density of postmenopausal women. J Bone Miner Res 1993;8:931–6. Prince R, Devine A, Dick I, et al. The effects of calcium supplementation (milk powder or tablets) and exercise on bone density in postmenopausal women. J Bone Miner Res 1995;10:1068–75. Kerr D, Morton A, Dick J, Prince R. Exercise effects on bone mass in postmenopausal women are site-specific and loaddependent. J Bone Miner Res 1996;11:218–25. Nelson ME, Fiatorone MA, Morganti CM, Trice I, Greenberg RA, Evans WJ. Effects of high-intensity strength training on multiple risk factors for osteoporotic fractures: a randomized controlled trial. JAMA 1994;272:1909–14. Notelovitz M, Martin D, Tesar R. Estrogen therapy and variableresistance weight training increase bone mineral in surgically menopausal women. J Bone Miner Res 1991;6:583–90. Pruitt LA, Taaffe DR, Marcus R. Effects of a one-year highintensity versus low-intensity resistance training program on bone mineral density in older women. J Bone Miner Res 1995;10:1788–95. Sinaki M, Wahner HW, Offord KP, Hodgson SF. Efficacy of nonloading exercises in prevention of vertebral bone loss in postmenopausal women: a controlled trail. Mayo Clin Proc 1989;64:762–9. Bassey EJ, Ramsdale SJ. Increase in femoral bone density in young women following high-impact exercise. Osteoporos Int 1994;4:72–5. Friedlander AL, Genant HK, Sadowsky S, Byl NN, Glu¨er C-C. A two-year program of aerobics and weight training enhances bone mineral density of young women. J Bone Miner Res 1995;10:574–85. Heinonen A, Kannus P, Sieva¨nen H, Oja P, et al. Randomised controlled trial of effect of high-impact exercise on selected risk factors for osteoporotic fractures. Lancet 1996;348:1343–7. Snow-Harter C, Bouxsein ML, Lewis BT, Carter DR, Marcus R. Effects of resistance and endurance exercise on bone mineral status of young women: a randomized exercise intervention trial. J Bone Miner Res 1992;7:761–9. Bloomfield SA, Williams NI, Lamb DR, Jackson RD. Non-

12

88. 89.

90. 91.

92. 93. 94. 95.

96.

97. 98.

99.

I. Wolff et al. weightbearing exercise may increase lumbar spine bone mineral density in healthy postmenopausal women. Am J Phys Med Rehabil 1993;72:204–9. Caplan GA, Ward JA, Lord SR. The benefits of exercise in postmenopausal women. Aust J Public Health 1993;17:23–6. Dalsky GP, Stocke KS, Ehsani AA, Slatopolsky E, Lee WC, Birge SJ. Weight-bearing exercise training and lumbar bone mineral content in postmenopausal women. Ann Intern Med 1988;108:824–8. Kohrt WM, Snead DB, Slatopolsky E, Birge SJ. Additive effects of weight-bearing exercise and estrogen on bone mineral density in older women. J Bone Miner Res 1995;10:1303–11. Nelson ME, Fisher EC, Dilmanian FA, Dallal GE, Evans WJ. A 1-y walking program and increased dietary calcium in postmenopausal women: effects on bone. Am J Clin Nutr 1991;53:1304–11. Pruitt LA, Jackson RD, Bartels RL, Lehnhard HJ. Weighttraining effects on bone mineral density in early postmenopausal women. J Bone Miner Res 1992;7:179–85. Gleeson PB, Protas EJ, LeBlanc AD, Schneider VS, Evans HJ. Effects of weight lifting on bone mineral density in premenopausal women. J Bone Miner Res 1990;5:153–8. Rockwell JC, Sorensen AM, Bakker S, et al. Weight training decreases vertebral bone density in premenopausal women: a prospective study. J Clin Endocrinol Metab 1990;71:988–93. Vuori I, Heinonen A, Sieva¨nen A, Kannus P, Pasanen M, Oja P. Effects of unilateral strength training and detraining on bone mineral density and content in young women: a study of mechanical loading and deloading on human bones. Calcif Tissue Int 1994;55:59-67. Walker GA, MacHannaford JC. A meta-analysis of randomized, double-blind, placebo-controlled studies of the effect of buflomedil on intermittent claudication. Fundam Clin Pharmacol 1995;9:387–94. Lanyon LE. Functional strain as a determinant for bone remodeling. Calcif Tissue Int 1984;36:S56-61. Lanyon LE. Using functional loading to influence bone mass and architecture: objectives, mechanisms, and relationship with estrogen of the mechanically adaptive process in bone. Bone 1996;18(Suppl):S37–43. O’Connor JA, Lanyon LE. The influence of strain rate on adaptive bone remodelling. J Biomechanics 1982;15:767–81.

100. Rubin CT, Lanyon LE. Regulation of bone mass by mechanical strain magnitude. Calcif Tissue Int 1985;37:411–7. 101. Rubin CT, Lanyon LE. Regulation of bone formation by applied dynamic loads. J Bone Joint Surg Am 1984;66:397–402. 102. Tommerup LJ, Raab DM, Crenshaw TD, Smith EL. Does weight-bearing exercise affect non-weight-bearing bone? J Bone Miner Res 1993;8:1053–8. 103. Specker BL. Evidence for an interaction between calcium intake and physical activity on changes in bone mineral density. J Bone Miner Res 1996;11:1539–44. 104. Cummings SR, Nevitt MC, Browner WS, et al. Risk factors for hip fracture in white women. N Engl J Med 1995;332:767–73. 105. Khosla S, Atkinson EJ, Riggs BL, Melton LJ. Relationship between body composition and bone mass in women. J Bone Miner Res 1996;11:857–63. 106. Trevisan C, Ortolani S, Bianchi ML, et al. Age, time since menopause, and body parameters as determinants of female spinal bone mass: a mathematical model. Calcif Tissue Int 1991;49:1–5. 107. Altman DG. Practical statistics for medical research. London: Chapman & Hall, 1992:455–9. 108. Baran DT. Magnitude and determinants of premenopausal bone loss. Osteoporos Int 1994;Suppl1:S31–4. 109. Riggs BL, Wahner HW, Melton LJ, Richelson LS, Judd HL, Offord KP. Rates of bone loss in the appendicular and axial skeletons of women: evidence of substantial bone loss before menopause. J Clin Invest 1986;77:1487–91. 110. Sowers MR, Galuska DA. Epidemiology of bone mass in premenopausal women. Epidemiol Rev 1993;15:374–98. 111. Welten DC, Kemper HCG, Post GB, van Staveren A. A metaanalysis of the effect of calcium intake on bone mass in young and middle aged females and males. J Nutr 1995;125:2802–13. 112. Cumming RG. Calcium intake and bone mass: a quantitative review of the evidence. Calcif Tissue Int 1990;47:194–201. 113. Whitcroft SIJ, Stevenson JC. Hormone replacement therapy: risks and benefits. Clin Endocrinol 1992;36:15–20. 114. Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med 1988;319:1701–7. 115. Joakimsen RM, Magnus JH, Fonnebo V. Physical activity and predispositions for hip fractures: a review. Osteoporos Int 1997;7:503–13.

Received for publication 15 September 1997 Accepted in revised form 27 April 1998