Abstract
Fracture probabilities derived from the original FRAX model for Brazil were compared to those from an updated model based on more recent regional estimates of the incidence of hip fracture. Fracture probabilities were consistently lower in the updated FRAX model. Despite large differences between models, differences in the rank order of fracture probabilities were minimal. OBJECTIVE: Recent epidemiological data indicate that the risk of hip fracture in Brazil is lower than that used to create the original FRAX model. This paper describes the epidemiology of hip fracture in Brazil and the synthesis of an updated FRAX model with the aim of comparing this new model with the original model. METHODS: Hip fracture rates from three cities in three regions were combined, weighted by the population of each region. For other major fractures, incidence rates for Brazil were estimated using Swedish ratios for hip to other major osteoporotic fracture (humerus, forearm or clinical vertebral fractures). Mortality estimates were taken from the UN. RESULTS: Compared to the original FRAX model, the updated model gave lower 10-year fracture probabilities in men and women at all ages. Notwithstanding, there was a very close correlation in fracture probabilities between the original and updated models (r > 0.99) so that the revisions had little impact on the rank order of risk. CONCLUSION: The disparities between the original and updated FRAX models indicate the importance of updating country-specific FRAX models with the advent of significant changes in fracture epidemiology.