Epidemiology of fractures in Armenia: development of a country-specific FRAX model and comparison to its surrogate

Fracture probabilities derived from the surrogate FRAX model for Armenia were compared to those from the model based on regional estimates of the incidence of hip fracture. Disparities between the surrogate and authentic FRAX models indicate the importance of developing country-specific FRAX models. Despite large differences between models, differences in the rank order of fracture probabilities were minimal. OBJECTIVE Armenia has relied on a surrogate FRAX model based on the fracture epidemiology of Romania. This paper describes the epidemiology of fragility fractures in Armenia used to create an Armenia-specific FRAX model with an aim of comparing this new model with the surrogate model. METHODS We carried out a population-based study in two regions of Armenia (Ararat and Vayots Dzor representing approximately 11% of the country's population). We aimed to identify all low-energy fractures: retrospectively from hospital registers in 2011-2012 and prospectively in 2013 with the inclusion of primary care sources. RESULTS The differences in incidence between the surveys with and without data from primary care suggested that 44% of patients sustaining a hip fracture did not receive specialized medical care. A similar proportion of forearm and humeral fractures did not come to hospital attention (48 and 49%, respectively). Only 57.7% of patients sustaining a hip fracture were hospitalized. In 2013, hip fracture incidence at the age of 50 years or more was 201/100,000 for women and 136/100,000 for men, and age- and sex-specific rates were incorporated into the new "authentic" FRAX model for Armenia. Compared to the surrogate model, the authentic model gave lower 10-year fracture probabilities in men and women aged less than 70 years but substantially higher above this age. Notwithstanding, there were very close correlations in fracture probabilities between the surrogate and authentic models (> 0.99) so that the revisions had little impact on the rank order of risk. CONCLUSION A substantial proportion of major osteoporotic fractures in Armenia do not come to hospital attention. The disparities between surrogate and authentic FRAX models indicate the importance of developing country-specific FRAX models. Despite large differences between models, differences in the rank order of fracture probabilities were minimal.