Publications

Abstract

Objectives: There are two commonly used fracture risk prediction tools FRAX�and Garvan Fracture RiskCalculator (GARVAN-FRC). The objective of this study was to investigate the utility of these tools in dailypractice.Study design: A prospective population-based 5-year follow-up study was conducted in ten generalpractice centres in the Netherlands. For the analyses, the FRAX�and GARVAN-FRC 10-year absoluterisks (FRAX�does not have 5-year risk prediction) for all fractures were used.Results: Among 506 postmenopausal women aged ?60 years (mean age: 67.8 � 5.8 years), 48 (9.5%)sustained a fracture during follow-up. Both tools, using BMD values, distinguish between women whodid and did not fracture (10.2% vs. 6.8%, respectively for FRAX�and 32.4% vs. 39.1%, respectively forGARVAN-FRC, p < 0.0001) at group level. However, only 8.9% of those who sustained a fracture had anestimated fracture risk ?20% using FRAX�compared with 53.3% using GARVAN-FRC. Although bothunderestimated the observed fracture risk, the GARVAN-FRC performed significantly better for womenwho sustained a fracture (higher sensitivity) and FRAX�for women who did not sustain a fracture (higherspecificity). Similar results were obtained using age related cut off points.Conclusions: The discriminant value of both models is at least as good as models used in other medicalconditions; hence they can be used to communicate the fracture risk to patients. However, given differ-ences in the estimated risks between FRAX�and GARVAN-FRC, the significance of the absolute