Fracture probability is predictive of fall-associated hospitalization: the Manitoba BMD Registry

The current version of FRAX® does not consider prior falls as a primary input variable. Limited data suggest that greater FRAX-derived fracture probability is associated with incident falls at least in elderly men. Our aim was to examine the association of FRAX-derived fracture probability with the risk for subsequent fall-associated hospitalizations. We identified individuals aged 40 years or older at the time of baseline DXA with FRAX-derived major osteoporotic fracture (MOF) probability assessed through the Manitoba BMD Program. We used linkage with population-based data to identify subsequent hospitalization that included a fall diagnosis code. Sensitivity analyses examined fall-associated hospitalizations unrelated to a concurrent fracture (diagnosed within 30 days). Cox regression was used to estimate hazard ratios (HR) for time to fall-associated hospitalization according to MOF probability without and with BMD (per SD increase and also for individual variables in the FRAX tool). The study comprised 88,684 individuals, mean age 64.6 years, 89.5% female. During mean 8.6 years observation (total 759,963 person-years), 9715 (11.0%) individuals experienced a fall-associated hospitalization; of these, 3363 (3.8%) were unrelated to concurrent fracture. Every SD increase in MOF fracture probability was strongly associated with fall-associated hospitalization (without BMD HR 2.47, 95%CI 2.41-2.52; with BMD HR 2.48, 95%CI 2.43-2.53). No significant interaction was seen between fracture probability and follow-up time (p=0.516). Findings were similar when restricted to individuals with falls unrelated to concurrent fracture, when adjusted for previous fall-associated hospitalization (last 3 years) or self-reported fall (last 12 months), when restricted to individuals without previous falls, and for FRAX-derived hip fracture probability. All FRAX variables except for parental hip fracture showed a positive relationship with fall-associated hospitalization. In conclusion, FRAX-derived fracture probability is strongly associated with risk for future fall-associated hospitalization, including falls unrelated to concurrent fracture.