Contributions of bone density and structure to fracture risk assessment in men and women.

Although bone mineral density (BMD) is a strong predictor of fractures, it is only a surrogate for bone strength. Bone structural parameters can now be measured on BMD scans, but it is unclear whether they would be more useful for risk assessment. We measured structural parameters using the Hip Structural Analysis Program and evaluated their association, compared with standard hip BMD, with fracture risk in a population-based sample of 213 postmenopausal women and 200 men > or =50 years of age. Altogether, 38% of the women and 27% of the men had experienced a fracture due to moderate trauma (half involved hip, spine or distal forearm), while 23% and 36%, respectively, had a previous fracture due to severe trauma. In logistic regression analyses adjusted for age, the hip BMD and structural parameters were all associated with moderate trauma fractures generally, and osteoporotic fractures specifically, in women, but the best predictor in a multivariate model was femoral neck BMD (odds ratio [OR], 2.8; 95% confidence interval [CI], 1.9-4.0). BMD and the structural parameters were strongly correlated, however, and could be interchanged with little reduction in predictive power. These variables were less predictive of moderate trauma fractures in men. The best model included age (OR per 10 years, 1.5; 95% CI, 1.1-2.1), femoral neck section modulus (OR, 1.6; 95% CI, 1.1-2.5) and intertrochanteric buckling ratio (OR, 1.6; 95% CI, 1.3-2.0). Correction for body size did not alter these relationships. Fractures due to severe trauma were best predicted by structural parameters: in women, femoral neck buckling ratio (OR, 1.2; 95% CI, 1.04-1.5) and, in men, intertrochanteric buckling ratio (OR, 1.4; 95% CI, 1.2-1.6). These data suggest that selected structural variables as assessed by dual-energy X-ray absorptiometry would be as good as standard BMD measurements for predicting fracture risk. Because of the strong correlations, however, some judgment can be used in selecting the variables easiest to measure.