Fracture risk prediction using FRAX®: a 10-year follow-up survey of the Japanese Population-Based Osteoporosis (JPOS) Cohort Study

Abstract

We evaluated the predictive ability of FRAX® in a cohort of 815 Japanese women. The observed 10-year fracture rate did not differ significantly from that predicted by FRAX®. The predictive ability of FRAX® without femoral neck bone mineral density (BMD) was similar to that with femoral neck BMD. We evaluated the ability of the Japanese version of FRAX®, a World Health Organization fracture risk assessment tool, to predict the 10-year probability of osteoporotic fracture. Self-reported major osteoporotic fracture (N = 43) and hip fracture (N = 4) events were ascertained in the 10-year follow-up survey of the Japanese Population-Based Osteoporosis Cohort Study. Participants were 815 women aged 40-74 years at the baseline survey. Receiver operating characteristic curve analysis compared FRAX® with multiple logistic models based on age, body weight, and femoral neck BMD. The number of observed major osteoporotic or hip fracture events did not differ significantly from the number of events predicted by the FRAX® model (with or without BMD). The area under the curve (AUC) value for FRAX® with BMD for predicting major osteoporotic fractures was similar to that of a logistic model with age, body weight, and BMD (0.69 vs. 0.71, respectively; p = 0.198); the AUC of FRAX® with BMD for predicting hip fractures was similar to that of a model based on age and BMD (0.88 vs. 0.89, respectively; p = 0.164). The AUCs of FRAX® without BMD for predicting major osteoporotic and hip fractures were similar to those with BMD (0.69 vs. 0.67, respectively; p = 0.121; 0.88 vs. 0.86, respectively; p = 0.445). The Japanese version of FRAX® without BMD estimated the 10-year probability of osteoporotic fracture in this population with few clinical risk factors as similar to that of FRAX® with BMD.