Background: Low bone mineral density (BMD) at the lumbar spine is a major public health problem among postmenopausal women. We conducted a meta-analysis of individual patient data (IPD) to examine the effects of exercise on lumbar spine BMD in postmenopausal women.
Methods: IPD were requested from a previously developed database of summary means from randomized and nonrandomized trials dealing with the effects of exercise on BMD. Two-way analysis of variance tests with pairwise comparisons (p < or =.05) and 95% confidence intervals (CIs) were used to determine the statistical significance for changes in lumbar spine BMD.
Results: Across 13 trials that included 699 subjects (355 exercise, 344 control), a statistically significant interaction was found between test and group (F = 15.232, p =.000). Pairwise comparisons (Bonferroni t tests) revealed a statistically significant increase in final minus initial BMD for the exercise group ( +/- SD = 0.005 +/- 0.043 g/cm(2), t = 2.46, p =.014, 95% CI = 0.001-0.009) and a statistically significant decrease in final minus initial BMD for the control group ( +/- SD = -0.007 +/- 0.045 g/cm(2), t = -3.051, p =.002, 95% CI = -0.012--0.002). Changes were equivalent to an approximate 2% benefit in lumbar spine BMD (exercise, +1%, control, -1%).
Conclusions: The results of this IPD meta-analysis suggest that exercise helps to improve and maintain lumbar spine BMD in postmenopausal women.