Background The development of FRAX ® has been useful to simplify the fracture risk assessment of patients, although several studies have shown that this tool underestimates the real risk of fracture in Spanish population.

Objectives To develop an accurate model of fracture risk assessment based on FRAX ®, using a cohort of Spanish women followed over a 10 year period and to establish the best thresholds for indication of DXA and indication of treatment (high-risk patients).

Methods Longitudinal multicenter study. Women between 40-90 years from FRIDEX cohort (35,000 people in the Barcelona with DXA and extensive questionnaire on risk factors for fracture at baseline), who had been followed during a 10 year period and who had not received any treatment for osteoporosis were selected. New self-reported osteoporotic fractures contrasted with electronic records or clinical reports were collected.The risk of major osteoporotic fracture for Spanish population based on FRAX was assessed in all women. We calculated the Area Under the ROC curve (AUC) for both DXA and 10-year risk of major fracture by FRAX.

Results We selected 816 women, mean age 56.8±8.2 years. After 10 years, 76 women (9.3%) had suffered 95 osteoporotic fractures, of which 49 (6%) were major (15 hip, 4 vertebral, 13 humerus and 17 distal radio). Women who suffered a major osteoporotic fracture during the follow-up were significantly older (p<0.001), had more previous fractures (p<0.001), had suffered more falls (p=0.016), and had a higher prevalence of osteoporosis by DXA (60% vs 15%) (p<0.001). The AUC-ROC for predicting future fracture was better for FRAX ® major fracture without and with BMD (0.736 and 0.733 respectively) than for DXA using T-score of femoral neck (0.697). The FRIDEX model was created selecting the best cut-off point for the 10-year absolute risk of major fracture by FRAX without BMD (Table 1). This model classifies women as a low risk (10-year risk or major fracture <5%), intermediate risk (5-7.5%) and high risk (>7.5%). In women with intermediate risk the model reassess the probability of fracture including the T-score of femoral neck by DXA in the FRAX tool and classify these patients as low or high risk for fracture. The application of this model could save 82% of DXA scans and avoid up to 31% of unnecessary treatments.

Conclusions The FRIDEX model, compared with DXA based model, has a better discriminative ability to detect women who will suffer osteoporotic fractures over a 10-year period. This model could save unnecessary DXA and osteoporotic treatments.

Disclosure of Interest R. Azagra Grant/Research support from: Sponsored by FEDER, IS Carlos III Grant. Spanish Ministry of Science., E. Casado: None Declared, G. Encabo: None Declared, A. Aguyé: None Declared, J. Pujol-Salud: None Declared, J. C. Martín-Sánchez: None Declared, E. Gené: None Declared, M. Zwart: None Declared, M. Iglesias: None Declared, F. Lόpez-Expόsito: None Declared, G. Roca: None Declared, S. Güell: None Declared