Vegan fractures in EPIC-Oxford: calcium was not the whole explanation
A prospective cohort found higher fracture rates in vegans, but it did not prove the cause or test a remedy.

Illustrative food image, not a photograph of the study intervention.
Fractures tracked over nearly eighteen years
EPIC-Oxford followed diet groups for an average of 17.6 years, identifying fractures through hospital records or death certificates. Baseline analyses of total fractures included 29,380 meat eaters, 8,037 fish eaters, 15,499 vegetarians and 1,982 vegans. Dietary information was collected in 1993–2001 and again around 2010. The researchers recorded 3,941 total fractures, including 945 hip fractures. This was a prospective observational study, not a trial assigning people to diets. Its strength is following actual fracture outcomes over time; its limitation is that statistical comparisons between self-selected diet groups cannot establish why the groups experienced different rates.
Tong TYN, Appleby PN, Armstrong MEG, Fensom GK et al.. BMC medicine, 2020. PMID 33222682 · DOI
Put the hip result in absolute terms
After adjustment for socioeconomic factors, lifestyle confounders and body mass index, the vegan group had a hip-fracture hazard ratio of 2.31 compared with meat eaters, with a 95% confidence interval of 1.66–3.22. The corresponding estimated rate difference was 14.9 additional cases per 1,000 people over ten years, with an interval of 7.9–24.5. Both presentations matter: the hazard ratio describes a relative comparison, whereas the rate difference conveys an absolute scale. Neither means that most vegans fractured a hip. Fish eaters and vegetarians also had higher hip-fracture estimates, though their reported absolute differences were smaller, at 2.9 additional cases per 1,000 over ten years.
Tong TYN, Appleby PN, Armstrong MEG, Fensom GK et al.. BMC medicine, 2020. PMID 33222682 · DOI
Calcium adjustment did not erase the association
The significant associations were generally stronger without adjustment for body mass index. Further adjustment for dietary calcium and/or total protein slightly attenuated them, but they remained significant. This is why the study cannot be reduced to a claim that calcium alone explains the observed vegan fracture difference. Statistical adjustment is also not an intervention: including calcium intake in a model does not test whether changing calcium intake prevents fractures. The abstract does not establish the mechanism behind the association or identify a menu that eliminates it. Its conclusion calls for further research into bone health in vegans rather than declaring the causal question settled.
Tong TYN, Appleby PN, Armstrong MEG, Fensom GK et al.. BMC medicine, 2020. PMID 33222682 · DOI
Not every fracture site showed a difference
Vegans also had higher estimates for total fractures, with a hazard ratio of 1.43, leg fractures at 2.05 and other main sites at 1.59 compared with meat eaters. Those other sites comprised clavicle, rib and vertebral fractures. But the abstract reports no significant diet-group differences for wrist or ankle fractures with or without body-mass-index adjustment, and none for arm fractures after that adjustment. A broad statement that every bone was at higher risk would therefore misrepresent the results. The honest framing preserves both findings: a concerning association at several sites, especially the hip, alongside null comparisons at others and unresolved causality.
Tong TYN, Appleby PN, Armstrong MEG, Fensom GK et al.. BMC medicine, 2020. PMID 33222682 · DOI