To use the meta-analytical approach to examine the effects of exercise on bone mineral density (BMD) in men.

MEDLINE, Current Contents, SPORTDiscus and Dissertation Abstracts International were searched from January 1966 to December 1998, using the following search terms alone or in various combinations: 'bone', 'exercise', 'physical activity', 'men', 'males', 'physical fitness', 'fitness' and 'osteoporosis'. The reference lists from original and review articles were also reviewed, and selected journals were handsearched. Studies published in foreign language journals, abstracts and conference papers were excluded.

Study designs of evaluations included in the review

Randomised controlled trials (RCTs) or non-randomised trials that included a comparative control group, lasting greater than or equal to 16 weeks.

Specific interventions included in the review

Exercise was compared with no exercise. The exercise protocols included sprinting, jumping, lumbar extension exercises, upper and lower body resistance, rowing, weight training, running, walking, jogging, swimming, stationary cycling and marathon training. Four studies reported that participants were taking calcium supplements.

Participants included in the review

Adult men with a mean study age of greater than or equal to 18 years. The participants in the intervention groups included power or endurance athletes, college oarsmen, runners, heart transplant patients, previously sedentary men or overweight males. The comparison groups included 'age and gender matched controls', heart transplant patients, previously sedentary men or overweight males. The participants were aged from 17 to 72 years.

Outcomes assessed in the review

The primary outcomes was changes in BMD. These were assessed at various sites using dual-energy X-ray absorptiometry, quantitative computed tomography or single-photon absorptiometry. The sites examined included the upper limb, lumbar spine (L1-L4 or L2-L3), femur (femoral neck, greater trochanter, Ward's triangle), tibia/fibula, mid-radius, total body, and/or os calcis.

The secondary outcomes examined were changes in body weight, the body mass index, the percentage of body fat, the lean body mass, the maximum oxygen consumption, the resting heart rate, and calcium intake.

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 reviewers performed the selection.

Validity was assessed using the 3-item, 5-point scale of Jadad et al. (see Other Publications of Related Interest no.1). The authors do not state how the papers were assessed for validity, or how many of the reviewers performed the validity assessment.

Two reviewers independently extracted the data using coding sheets. The authors met and reviewed the data for accuracy and consistency. Any disagreements were resolved by consensus. The major categories of variables coded included the study characteristics, the physical characteristics of the participants, and the primary and secondary outcomes. The following data were tabulated: study reference; study design; details of the participants, such as sample size, physical characteristics and mean age; the duration and type of exercise intervention; and BMD assessment method and site of measurement.

How were the studies combined?

A standardised difference was used as the measure of the effect size (ES). The ES was then corrected for small-sample bias. On account of the small sample sizes (particularly for subgroup analyses), bootstrap re-sampling (5,000 iterations) was used to generate the 95% bootstrap confidence intervals (BCIs) around the mean ES changes for BMD.

Due to the significant heterogeneity present, the authors used a random-effects model to combine the studies; the significance level was less than or equal to 0.05. Publication bias was examined using Kendall's tau statistic (see Other Publications of Related Interest no.2), and by creating a funnel plot.

How were differences between studies investigated?

The heterogeneity of ES changes in BMD was examined using the Q statistic. Subgroup analyses were performed using procedures similar to analysis of variance for the meta-analysis (see Other Publications of Related Interest no.3). The subgroup analyses were conducted to investigate: those BMD sites assessed, which were specific to the sites loaded during exercise; the age of the participants; site-specific results for older participants in relation to a specific assessment location (femur, lumbar, os calcis); and study design. Sensitivity was examined by conducting analyses with each study deleted from the model. Regression analyses were also undertaken to examine correlations between ES changes in BMD and the initial BMD or duration of training.

Eight studies were included in the review: 2 RCTs and 6 non-randomised controlled trials. The authors reported that there were 225 participants: 135 in the exercise groups and 90 in the controls. However, there is a discrepancy between the values reported in the text and those tabulated. Based on the data from a table, there were 249 participants: 153 in the exercise groups and 96 in the controls. For the RCTs, there were 29 and 27 participants in the intervention and control groups, respectively. The sample sizes reported by the authors appear to consider withdrawals or drop-outs.

Primary outcomes.

There was statistically-significant heterogeneity between the studies (Q=45.01, p=0.008). A random-effects model demonstrated that overall ES changes were not significant. ES changes were equivalent to an improvement (exercise minus control) of 2%: there was a 1.6% increase in the exercising participants and a 0.4% decrease in the controls. Subgroup analyses.

When the BMD sites assessed were specific to the sites loaded during exercise, increases of around 2.6% were found: 2.1% in the exercisers and 0.5% in the controls. These results were statistically significant (ES=0.213, 95% BCI: 0.007, 0.452).

Statistically-significant ES changes were found for older (greater than 31 years) men (ES=0.605, 95% BCI: 0.324, 1.032), but not for younger (less than 31 years) men. The differences between the groups were statistically significant (Qb=5.89, p=0.04). Statistically-significant changes were also observed at the femur (ES=0.482, 95% BCI: 0.270, 0.705), lumbar (ES=0.749, 95% BCI: 0.099, 1.327), and os calcis (ES=0.565, 95% BCI: 0.260, 1.048) sites. Statistically-significant changes were not observed between groups. The within-group ES increases in BMD were also found to be statistically significant when data were examined by study design. Statistically-significant increases were not observed between groups. No statistically-significant correlations were found between ES changes in BMD and the duration of training or initial BMD.

A number of secondary outcomes (e.g. body weight, body mass index, lean body mass) were also examined, but no statistical differences were observed. Study quality ranged from 0 to 3; only one study received a score of 3.

There was no quantitative evidence supporting publication bias (r=- 0.23, p=0.10), although the funnel plot analysis revealed some potential bias.

The results of this study suggested that site-specific exercise may help improve and maintain BMD at the femur, lumbar, and os calcis sites in older men. However, there were several factors that prevented any firm conclusions being drawn on the use of exercise for maintaining and/or improving BMD in men. These were the biological importance of the small changes observed for most outcomes, the quality of the studies, and the limited data pool.

The authors stated their review question clearly, and the inclusion and exclusion criteria were well defined, although some of the criteria were broad. There was a discrepancy in the number of participants included: the authors stated that 225 participants were included in the review, whereas 249 participants were reported in a table. The number of evaluable participants for each study should have been clearly presented.

The literature search was adequate, although foreign language journals, abstracts and conference papers were excluded. It is likely, therefore, that some studies may have been missed.

The validity of the included studies was systematically assessed. However, the results of this assessment were not presented for each study; probably because seven of the eight studies received relatively poor scores, which the authors reported rather vaguely as 'ranging from 0 to 2'. However, the authors discussed the poor quality of included studies, and made some useful methodological suggestions for future studies.

Some data was presented on the individual studies. Information on baseline measures of BMD for the intervention and control groups, and outcomes may have been useful. The authors suggested that their funnel plot was indicative of publication bias, although there was insufficient information provided to justify this.

Pooling may have been inappropriate as there appeared to be considerable heterogeneity between the small number of included studies. While the authors excluded the study of heart transplant patients from some of their analyses, less obvious heterogeneity may have remained. The authors stated that no firm conclusions regarding the use of exercise for maintaining and/or improving BMD in men could be made. This seemed appropriate given the current evidence available.

Practice: The authors state that if exercise is recommended, it should only be performed in conjunction with other types of non- pharmacological and/or pharmacological interventions.

Research: The authors state that there is a clear need for additional studies in this area.

US Department of Defence, Army Medical Research and Material Command Award number 17-98-1-8513.

1. Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ, Gavaghan DJ, et al. Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 1996;17:1-12. 2. Begg CB. Publication bias. In: Cooper H, Hedges LV, editors. The handbook of research synthesis. New York: Russell Sage; 1994. p. 399- 409. 3. Hedges LV, Olkin I. Statistical methods for meta-analysis. San Diego (CA): Academic Press; 1985.

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.