Confounders in the association between exercise and femur bone in postmenopausal women

Abundant animal and human evidence demonstrates that loading stimuli generate positive adaptive changes in bone, but effects of activity on bone mineral density (BMD) are often modest and frequently equivocal.

Physical activity effects on the femur would be better reflected in measurements of geometry than BMD.

Cross-sectional cohort study.

We used data from 6032 women of mixed ethnicity aged 50-79 yr who had dual-energy x-ray absorptiometry (DXA) scans of the total body and hip from the Women's Health Initiative observational study. Subjects were distributed in three ways: self-report categories included 1) tertiles of MET and 2) reported minutes per week walking for exercise. A third, more objective, category was based on tertile of lean body mass fraction (LMF) from DXA scans. Femur outcomes included conventional femoral neck and total hip BMD, bone mineral content and region area, and geometry measurements using the Hip Structure Analysis software. Outcomes were compared between activity groups using models adjusted for common confounders.

Adjusted bone measurements showed similar activity effects with all three grouping variables, but these were greater and more significant when evaluated by LMF tertile. Women in the highest LMF tertile had the widest femurs. Differences in section modulus between highest and lowest tertile of LMF were 50%-80% greater than the association with bone mineral content and two to three times that on BMD.

More active women in the Women's Health Initiative observational study had geometrically stronger femurs, although effects are underestimated, not apparent, or sometimes negative when using BMD as an outcome.

Exercise improves the strength of the femur largely by adding bone to the outer cortical surface; this improves resistance to bending, but because of the way DXA measurements are made, this may paradoxically reduce BMD.