The association between graded physical activity in postmenopausal British women, and the prevalence and incidence of hip and wrist fractures

Background: physical activity is promoted for older women as a means of maintaining health and avoiding falls and fractures. Findings relating physical activity of older women to risk of falls and fracture are contradictory. The association between level of physical activity and prevalent and incident hip and wrist fractures was examined in a large representative sample of postmenopausal British women. Methods: data from the British Women's Heart and Health Study, a cohort study of 4286 postmenopausal women aged 6079, from 23 UK towns were used. Information on physical activity, anthropometry, falls and hip and wrist fractures from baseline examination and questionnaire (1999-2001) and follow-up questionnaire (2007) were available. Cross-sectional baseline prevalence data were analysed using logistic regression and cohort incidence data using a Cox proportional hazards model examining the association of physical activity with fracture outcomes. Results: 3003 (70%) women, with complete baseline data, were studied. 13.6% had previously fractured a wrist and 1.3% a hip. Analyses unadjusted for confounders showed moderate protective associations between activity and fracture risk. After adjustment for confounders there was a weak trend towards fewer hip fractures (adjusted OR 0.13 [0.01, 1.18]) and more wrist fractures (adjusted OR 1.35 [0.76, 2.48]), amongst most active compared with inactive women. The crude incidence rate of wrist and/or hip fracture was 7.0 [5.9, 8.2] per 1000 person-years. No evidence was found for an association between physical activity and combined incident hip and/or wrist fracture (adjusted rate ratio inactive versus most active 1.69 [10.67, 4.24]). Conclusion: no clear associations between graded physical activity and hip/wrist fractures were seen but estimates were imprecise. Physical activities are heterogeneous and individual fracture types and mechanisms differ. Very large prospective observational studies are required to disentangle the precise effects of different activity patterns on different fracture types.