FRAX predicts incident falls in elderly men: findings from MrOs Sweden.
Harvey NC., Johansson H., Odén A., Karlsson MK., Rosengren BE., Ljunggren Ö., Cooper C., McCloskey E., Kanis JA., Ohlsson C., Mellström D.
UNLABELLED: Falls and fractures share several common risk factors. Although past falls is not included as an input variable in the FRAX calculator, we demonstrate that FRAX probability predicts risk of incident falls in the MrOs Sweden cohort. INTRODUCTION: Although not included in the FRAX® algorithm, it is possible that increased falls risk is partly dependent on other risk factors that are incorporated into FRAX. The aim of the present study was to determine whether fracture probability generated by FRAX might also predict risk of incident falls and the extent that a falls history would add value to FRAX. METHODS: We studied the relationship between FRAX probabilities and risk of falls in 1836 elderly men recruited to the MrOS study, a population-based prospective cohort of men from Sweden. Baseline data included falls history, clinical risk factors, bone mineral density (BMD) at femoral neck, and calculated FRAX probabilities. Incident falls were captured during an average of 1.8 years of follow-up. An extension of Poisson regression was used to investigate the relationship between FRAX, other risk variables, and the time-to-event hazard function of falls. All associations were adjusted for age and time since baseline. RESULTS: At enrolment, 15.5 % of the men had fallen during the preceding 12 months (past falls) and 39 % experienced one or more falls during follow-up (incident falls). The risk of incident falls increased with increasing FRAX probabilities at baseline (hazard ratio (HR) per standard deviation (SD), 1.16; 95 % confidence interval (95%CI), 1.06 to 1.26). The association between incident falls and FRAX probability remained after adjustment for past falls (HR per SD, 1.12; 95%CI, 1.03 to 1.22). High compared with low baseline FRAX score (>15 vs <15 % probability of major osteoporotic fracture) was strongly predictive of increased falls risk (HR, 1.64; 95%CI, 1.36 to 1.97) and remained stable with time. Whereas past falls were a significant predictor of incident falls (HR, 2.75; 95%CI, 2.32 to 3.25), even after adjustment for FRAX, the hazard ratio decreased markedly with increasing follow-up time. CONCLUSIONS: Although falls are not included as an input variable, FRAX captures a component of risk for future falls and outperforms falls history with an extended follow-up time.