To measure the effect of physical activity on the bone mass of healthy sedentary postmenopausal women.

MEDLINE was searched from January 1966 to May 1996 for articles published in French or English, using the keywords 'osteoporosis' combined with either 'weight bearing', 'bone loading' 'physical activity' or 'exercise'. The bibliographies of relevant articles were also examined. Only full-length articles were selected for the review. Abstracts from conference proceedings were excluded.

Study designs of evaluations included in the review

Prospective intervention trials, which included a comparison group (randomised or non-randomised) were included.

Specific interventions included in the review

Exercise programmes of various duration and intensity. These included: walking; jogging; treadmill; stair climbing; cycling; squeezing a tennis ball; weight-bearing exercise for the back; aerobics and musculation; high-impact exercise; low-impact exercise; climbing up and down a bench; weight lifting with gym equipment; use of free weights; flexibility and calisthenic exercises; and stretching exercises. The length of the physical activity programmes varied from 1.5 months to 5 years. Studies evaluating a physical activity programme concomitantly with any other treatment or drug therapy, such as oestrogen, vitamin D or calcium, were excluded.

Participants included in the review

Healthy sedentary postmenopausal women of at least 50 years of age (mean age: 49 to 84) who were free of symptomatic osteoporosis at the time of study entry. One study also included women who were veteran swimmers.

Outcomes assessed in the review

Bone parameters such as bone mineral content and bone mineral density were measured at different anatomical sites.

How were decisions on the relevance of primary studies made?

The authors do not state how the papers were selected for the review, or how many of the authors performed the selection.

A modified version of the scoring system of Chalmers et al. (see Other Publications of Related Interest no.1) was used to assess validity. The scoring system has a range from 0 to 100, and assigns scores according to the value of the protocol (60 points), statistical analysis (30 points), and presentation of results (10 points). The protocol was assessed with respect to the selection process, definition of the exercise programme, calculation of sample size, and reporting of compliance. The results section was scored according to the baseline data analysis, appropriateness of the statistical tests, handling of withdrawals, consideration of side-effects, and presence of sensitivity analyses. Three investigators evaluated the quality of each study independently, and any disagreements were resolved through consensus.

Three investigators carried out the data extraction using standardised forms developed a priori. Two data extractors were blinded to the name of the authors, the journal title, year of publication, and the existence of financial support.

How were the studies combined?

The effect sizes for parameter-specific and anatomical-specific sites were calculated according to Hedges and Olkin (see Other Publications of Related Interest no.2), then combined using the random-effects model of DerSimonian and Laird (see Other Publications of Related Interest no.3). If a study was present more than once when combining effects (i.e. a study with two exercise groups), one effect size was randomly chosen for the calculation in order to keep the data independent of each other.

Multivariate analyses were performed to test the effect of the quality score, age, measuring instrument, design, and programme characteristics such as duration and intensity.

How were differences between studies investigated?

The homogeneity of effects between studies was assessed using a chi-squared test.

Eighteen studies (846 participants) were included: 5 randomised controlled trials, 11 non-randomised controlled trials, and 2 matched studies.

The mean overall quality score of the studies was 47 (range: 27 to 76). The main weaknesses of the studies were: a lack of information regarding the number of participants screened; a priori estimates of the sample size; and the estimated power for non significant results.

The heterogeneity statistic was found to be significant in all cases.

Overall, physical activity had no statistically-significant effect on any of the sites measured. A sensitivity analysis conducted on studies published after 1991 showed a significant effect of physical activity on the bone mineral density at the L2-4 level of the lumbar column (effect size 0.8745, p<0.05). No effect could be seen, however, on forearm and femoral bone mass. Sensitivity analyses were also performed, excluding the study with 49-year-olds, on studies having a quality score higher than 50 and on the overall effect of age and measuring instruments. Although no significant effect was observed in any of these analyses, the loss of information caused by reducing the number of studies may have also reduced the statistical power to detect such an effect.

Although only applied to a small number of studies, this meta-analysis suggested that exercise programmes in a population of postmenopausal women over 50 years of age are effective for preventing spinal bone mineral density loss at the L2-4 level. When taking all studies published between 1966 and 1996 into account, physical activity was not found to have a beneficial effect on bone loss after the menopause.

This was a thorough review that was conducted in a rigorous manner using specific inclusion criteria, and with a clear validity assessment of the included studies. However, the literature search was somewhat limited in that only one electronic database was searched, and no attempt was made to trace any unpublished data. The authors' conclusions appear to follow from the results.

Natural Sciences and Engineering Research Council of Canada.

1. Chalmers TC, Smith H, Blackburn B, Silverman B, Schroeder B, Reitman D, et al. A method for assessing the quality of a randomized control trial. Control Clin Trials 1981;2:31-49. 2. Hedges L, Olkin I. Statistical methods for meta-analysis. Florida: Academic Press; 1985. 3. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177-88.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.