FRACTURE RISK ASSESSMENT WITH FRAX USING REAL-WORLD DATA: A POPULATION-BASED COHORT FROM ISRAEL.

Abstract

The predictive value of the WHO fracture risk assessment tool was evaluated using real-world community data. A population-based cohort of 141,320 women aged 50-90 years old (median age 58, inter quartile range (IQR): 54-67) at the year 2004 was extracted from the central database of a large Israeli healthcare services provider and insurer. Retrospective FRAX scores were calculated using computerized health records and compared to actual incident major osteoporotic fractures (MOF) during the following 10 years. A total of 6.9% and 2.2% MOF and hip fractures were expected vs. 13.5% and 2.9% observed. The area under receiver-operating characteristic curve (AUC) of FRAX without bone mineral density (BMD) was 0.65 (95% CI: 0.65, 0.66) for MOF and 0.82 (95% CI: 0.81, 0.82) for hip fracture. A total of 16,578 subjects had BMD data at index, and their risk estimates based solely on BMD exhibited lower predictive performance for MOF (AUC = 0.62 vs. 0.65, P value = 0.003) as well as hip fractures (AUC = 0.78 vs. 0.84, P value < 0.001) as compared with FRAX. FRAX using electronic health records provided reasonable discrimination despite some underestimation of the absolute risk of non-hip fractures. Integration with routine clinical systems could increase implementation in daily practice and improve risk detection, especially for patients without BMD.