Objectives Patients with knee osteoarthritis have an increase in bone mass but no corresponding decrease in risk of fracture. This study describes the rates of hip fracture in subjects with knee osteoarthritis before and after having a total knee replacement (TKR), compared with matched controls.

Methods A population-based prospective cohort study was conducted. The study population included, from the General Practice Research Database (UK), patients 40 years and older, undergoing TKR between 1986 and end-2006 for knee osteoarthritis as ‘cases’ (n=20 033). Five disease-free controls (n=100 165) were randomly selected, and matched for age, gender and practice. Hip fractures were ascertained using READ codes, and yearly rates of hip fracture and rate differences were calculated for the 5 years before and after surgery, using Poisson regression. Stratified analyses were performed by age and history of fracture.

Results Hip fracture rates were non-significantly reduced compared with controls before the operation. In the year after TKR, risk increased significantly (RR 1.58; 95% CI 1.14 to 2.19). Rates then declined to equal those of controls by 3 years, and continued decreasing until the end of follow-up; corresponding RR were not significant. The increased risk is greatest in younger ages and in those without previous fracture.

Conclusions The association between knee osteoarthritis and fractures is time-dependent, which may explain the current controversy in the literature. The association is also modified by TKR: subjects have a higher rate of hip fracture than matched controls after TKR, although the rates may eventually decrease.