Predictive value of FRAX for fracture in obese older women

Abstract

Recent studies indicate that obesity is not protective against fracture in postmenopausal women and increases the risk of fracture at some sites. Risk factors for fracture in obese women may differ from those in the nonobese. We aimed to compare the ability of FRAX with and without bone mineral density (BMD) to predict fractures in obese and nonobese older postmenopausal women who were participants in the Study of Osteoporotic Fractures. Data for FRAX clinical risk factors and femoral neck BMD were available in 6049 women, of whom 18.5% were obese. Hip fractures, major osteoporotic fractures, and any clinical fractures were ascertained during a mean follow-up period of 9.03 years. Receiving operator curve (ROC) analysis, model calibration, and decision curve analysis were used to compare fracture prediction in obese and nonobese women. ROC analysis revealed no significant differences between obese and nonobese women in fracture prediction by FRAX, with or without BMD. Predicted hip fracture risk was lower than observed risk in both groups of women, particularly when FRAX + BMD was used, but there was good calibration for FRAX + BMD in prediction of major osteoporotic fracture in both groups. Decision curve analysis demonstrated that both FRAX models were useful for hip fracture prediction in obese and nonobese women for threshold 10-year fracture probabilities in the range of 4% to 10%, although in obese women FRAX + BMD was superior to FRAX alone. For major osteoporotic fracture, both FRAX models were useful in both groups of women for threshold probabilities in the range of 10% to 30%. For all clinical fractures, the FRAX models were not useful at threshold probabilities below 30%. We conclude that FRAX is of value in predicting hip and major osteoporotic fractures in obese postmenopausal women, particularly when used with BMD.