Factors That Predict 1-Year Incident Hip and Non-Hip Fractures for Home Care Recipients: A Linked-Data Retrospective Cohort Study

Abstract

ObjectivesThe purpose of our study was to identify factors that predict 1-year incident hip and major osteoporotic non-hip fractures (ie, wrist, spine, pelvis, humerus) for home care recipients while accounting for the competing risk of death. DesignWe conducted a retrospective cohort study with linked population data. Setting and ParticipantsAll home care recipients in Ontario, Canada, receiving services for more than 6 months with an admission assessment between April 1, 2011, and March 31, 2015, were included. MethodsClinical data from the Resident Assessment Instrument Home Care were linked to fracture data from the Discharge Abstract Database and the National Acute Care Reporting System. Competing risk proportional hazard regressions using the Fine and Grey method were performed to model the association between potential risk factors and fracture. ResultsPrevious fall, previous fracture, cognitive impairment, unsteady gait, alcohol use, tobacco use, and Parkinson disease were consistently associated with all fracture types. Cognitive impairment (hazard ratio 2.09; 95% confidence interval 1.86–2.36) and wandering [1.66 (1.06–1.27)] were most predictive of hip fractures and being female [1.86 (1.76–1.98)] and experiencing a previous fracture [1.86 (1.76–1.98)] were most predictive of non-hip fractures. Risk factors unique to non-hip fractures as compared with hip fractures were locomotion ability outdoors and psychotropic medication use. Conclusions and ImplicationsOur results indicate that, in addition to typical fracture risk factors, home care recipients have unique characteristics that increase their risk. Fracture risk assessment tools and subsequent prevention strategies should be modified to accurately identify home care recipients at risk for imminent 1-year fracture.