Exercise lifts bone density at five of six sites

Bone density falls with age, and exercise is the main non-drug lever people are told to pull. What has never been clear is which kind of training works best, whether it works everywhere in the skeleton, and how much of it is needed. Bone responds locally to loading, so a programme that helps the spine may do little for the hip, and generic advice to exercise glosses over that. Putting a number on both the site-specific effect and the weekly dose would make guidance far more useful. This review set out to compare exercise types across six skeletal sites and test whether weekly exercise volume tracks bone change.

The authors searched four databases to March 2026 and included 162 randomised controlled trials covering 10,475 participants aged 18 to 81.6 years. Three analytical frameworks were applied: a three-level pairwise pooling with robust variance estimation, a Bayesian network analysis ranking exercise types across sites, and an exploratory dose-response model using restricted cubic splines fitted to weekly MET-minutes. Methodological quality was judged with ROBUST-RCT and GRADE, plus an additional certainty assessment specific to network comparisons. Outcomes were bone mineral density at the lumbar spine, femoral neck, Ward's triangle, trochanter, total hip and total body.

Significant pooled improvements appeared at five of the six sites: femoral neck (g 0.28), total body (g 0.26), Ward's triangle (g 0.23), lumbar spine (g 0.22) and trochanter (g 0.15). Certainty in the pairwise results ranged from high at the femoral neck to very low at the total hip. The network analysis produced site-specific rankings for different exercise types, but the credible ranges were often wide and certainty was limited by imprecision, disagreement between studies and sparse direct comparisons. The dose analysis suggested an overall peak near 1100 MET-min/week for lumbar spine density, with aerobic plus resistance training appearing to peak around 1800 MET-min/week, though both figures were sensitive to the model's boundaries.

This is a large consolidation of a field where individual trials are small and inconsistent, and it confirms the broad message that loading the skeleton helps across adulthood. What it complicates is the idea that one modality is clearly superior: the rankings exist but are too imprecise to be treated as instructions. The effect sizes are also a reminder that bone changes slowly and modestly with training rather than dramatically. Open questions include which loading patterns suit which sites, why the total hip evidence is so weak, and whether the suggested dose peaks survive better-designed trials.

This is a reason to be confident that regular exercise supports bone density, and equally a reason to distrust confident claims about the one best bone-building workout. The gains reported here were small to moderate and varied by skeletal site, so expectations should be calibrated accordingly. The weekly volume figures are exploratory, and the authors explicitly say they should not be read as a prescription. For most readers, the practical signal is that consistent loading across adulthood matters more than picking the theoretically optimal modality.