Quality and performance measures in bone densitometry. Part 2: fracture risk.

Abstract

This is part 2 of a core chapter of the forthcoming Report on Bone Densitometry commissioned by the International Commission on Radiation Units (ICRU). It is written with the aim to review definitions of quantities and units used in bone densitometry research and to describe parameters and methods that can be used to compare and standardize densitometric equipment and measurements. Part 2 of this chapter contains the section on fracture risk. Building on concepts of risk assessment, including risk ratios and odds ratios, we review statistical concepts commonly used in cross-sectional and prospective fracture studies. Performance measures are defined that allow a comparison of the ability of densitometry techniques to assess fracture risk. The methods of discriminant analysis, logistic regression, Poisson regression models, and the Cox proportional hazard model are presented and compared. In addition, statistical concepts to characterize risk for the individual patient are reviewed.