To examine the effects of aerobic exercise on lumbar spine bone mineral density in postmenopausal women.

MEDLINE was searched from January 1975 to December 1994, and Current Contents from January 1993 to December 1994, using the following keywords either alone or in combination: 'exercise', 'physical activity', ' physical fitness', 'bone', 'bone density', 'osteoporosis','lumbar', 'spine', 'women', 'female', 'postmenopausal' and 'menopausal'. Only journals published in the English language were searched. The author also checked the retrieved articles for additional studies, and used a funnel plot to test for publication bias.

Study designs of evaluations included in the review

Randomised or non-randomised trials where aerobic exercise was the primary intervention, with a comparative non-exercise group. The trial had to measure changes in lumbar spine bone density in postmenopausal women and report changes in lumbar spine bone mineral density.

Specific interventions included in the review

Aerobic exercise ranging from 15 to 50 to 60 minutes in duration, 1 to 4 times a week, over a period of approximately 28 to 80 weeks. This included stationary cycling, low- and high-impact aerobics, walking, jogging, stair-climbing, treadmill, step-ups, upper trunk movements, and water aerobics. The control group had no exercise.

Participants included in the review

Postmenopausal women were included.

Outcomes assessed in the review

Changes in lumbar spine bone mineral density were measured by dual-energy X-ray absorptiometry (3 studies), dual-photon absorptiometry (5 studies), and computed tomography scan (1 study). One study measured lumbar bone density at more than one site using both dual-photon absorptiometry and computed tomography scan.

How were decisions on the relevance of primary studies made?

One blinded author examined the methods and results sections separately under coded conditions.

The author did not state that they assessed validity.

The author used a coding sheet to extract up to 305 pieces of information for each study. The author was blinded to the author(s), journal and year of publication. The author coded the same studies twice, two weeks apart and used a codebook with detailed description of the information coded. The coding sheet was then modified until there was 100% agreement between both coding sessions.

The author also used the bootstrap technique (see Other Publications of Related Interest no.1) to increase the sample size (10,000 repeat samples using replacement) and to estimate the reliability of the change in treatment effect estimates. All estimates of the treatment effects were corrected for sample size using the procedures developed by Hedges and Olkin (see Other Publications of Related Interest no.2).

How were the studies combined?

The treatment effect measurements were combined using a fixed-effect model. Significance was set at a p-value less than or equal to 0.05.

Subgroup analyses were performed using a Mann-Whitney rank sum test to compare differences between changes in lumbar bone density when partitioned by study design, country in which the study was conducted, with or without calcium supplementation, and high- versus low- impact exercise. A Kruskall-Wallis analysis of variance on ranks was used to compare changes in lumbar bone mineral density when partitioned according to the method of assessment.

Confidence intervals were used to examine changes when certain groups were deleted, i.e. women receiving hormone replacement therapy, smokers and women who consumed alcohol.

Correlation analyses were performed using Spearman's rho tests to examine the relationship between changes in the following: lumbar bone mineral density, study characteristics and physical characteristics, lumbar bone mineral density assessment, training programme characteristics, and initial lumbar bone mineral density.

A Mann-Whitney rank sum test was used to compare differences between exercise and non-exercise groups in relation to drop-outs, initial age, body mass index, and maximum oxygen consumption.

Independent t-tests were performed to compare differences in relation to the following: the number of participants; years past menopause; initial height and levels of calcium intake; outcome changes in body mass index and maximum oxygen consumption; and initial and outcome changes in body weight.

How were differences between studies investigated?

Heterogeneity was examined using a box plot to identify outlier results.

Ten controlled trials with 330 participants (192 in the aerobic exercise group and 138 in the non-exercise group) were included. There were 4 randomised controlled trials and 6 non-randomised controlled trials.

Using the fixed-effect model, the bootstrap analysis showed that statistically-significant changes in lumbar spine bone mineral density occurred. Changes in lumbar bone mineral density, corrected for sample size, ranged from -2.41 to +8.12% (mean 2.83, standard deviation 3.26%, 95% confidence intervals: 1.15, 4.50) across all designs and categories, This overall average increase was caused primarily by the loss of lumbar spine bone mineral density in the non-exercise group, relative to the exercise group.

No statistically-significant differences or important relationships were reported for any of the subgroup or correlation analyses.

Although not statistically significant, there was a trend for changes in lumbar bone density to be positively correlated with initial calcium levels (correlation 0.62, P=0.067).

The funnel plot showed publication bias, with an absence of studies with small sizes.

The results suggested that aerobic exercise helps to maintain lumbar spine bone mineral density in postmenopausal women.

This was a reasonably complete systematic review. The author searched several databases and checked bibliographies for additional studies. However, since the search was limited to English language publications, it is unclear whether additional studies may have been missed.

The inclusion criteria were stated, but the author did not assess either the validity or the quality of the included studies. The data were extracted by the author alone, and although this task was repeated, it does not meet the criteria of independent selection or the extraction criteria.

The numbers of participants quoted in the text of the review did not match the total participant numbers in the tabulated data: 365 in the tables versus 330 quoted in the text.

The author used a very wide and appropriate range of statistical analyses to combine the data and to perform subgroup and correlation analyses. A bootstrap analysis was applied to increase the sample size, but it is unclear whether this method was appropriate since the use of randomisation was poor in the included studies (4 out of 10 studies were randomised). Heterogeneity was assessed but these results were not reported; this may also have affected the use of the bootstrap method.

The conclusions of the review, that aerobic exercise helps to maintain lumbar spine bone mineral density in postmenopausal women follows from the results. However, these conclusions should be viewed with caution because there was little transparency (implying bias) in the selection and extraction of the data, there was no quality review of the included studies, and no discussion of the effects of heterogeneity on the methods used in the analyses.

The author states that there is a need for additional well-designed studies on this topic before a final recommendation can be made regarding the effects of aerobic exercise on lumbar spine bone mineral density.

1. Effron B, Tibshirani R. An introduction to the bootstrap. London: Chapman and Hall; 1993. 2. Hedges LV, Olkin I. Statistical methods for meta-analysis. San Diego (CA): Academic Press; 1985. 3. Kelley G. Exercise and regional bone mineral density in postmenopausal women: a meta-analytic review of randomized trials. Am J Phys Med Rehabil 1998;77:76-87.

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.