Introduction We compared the relationship between body mass index (BMI) and physical activity for the risk of ankle, wrist and hip fractures in a large prospective study of postmenopausal women in the UK.

Methods In 1996–2001, women recruited into the Million Women Study completed a self-administered questionnaire asking about a range of health and lifestyle factors. Incident fractures were identified through self-report in a follow-up questionnaire completed on average 3.2 years after recruitment. RRs and CIs for each fracture site in women by BMI and physical activity at recruitment were calculated using Cox regression models, adjusted for socioeconomic status, and other factors.

Results Among 599 648 postmenopausal women, there were 5117 ankle fractures, 8564 wrist fractures, and 753 hip fractures. When compared to lean women (BMI<20.0 kg/m²), obese women (BMI≥30 kg/m²) had an increased risk of ankle fracture (RR=2.47; 95% CI 2.32 to 2.63), but a decreased risk of wrist fracture (RR=0.69; 95% CI 0.65 to 0.73) and hip fracture (RR=0.29; 95% CI 0.23 to 0.37). Physical activity had little influence on the risk of ankle or wrist fracture, but women who reported partaking in strenuous physical activities were at a lower risk of hip fracture than women who reported being never/rarely active (RR=0.65; 95% CI 0.56 to 0.76).

Conclusion BMI and physical activity have different effects on the incidence of fracture at different sites. While obese women are at increased risk of ankle fracture they are at lower risk of wrist and hip fracture. Physical activity has no marked influence on ankle and wrist fracture but is protective against hip fracture.