Comparative effects of densitometric and absolute fracture risk classification systems on projected intervention...

It is important to understand how the move to absolute fracture risk estimation will affect patient categorization. We retrospectively compared categorization systems in 17,053 women aged 50 yr and older from a large referral database of clinical bone mineral density (BMD) tests. Densitometric systems (femoral neck alone or minimum value from spine, total hip, femoral neck, and trochanter) were taken to indicate high risk based upon T-score -2.5 or lower. Ten-year absolute fracture risk of the hip, spine, wrist, and proximal humerus was estimated from T-score and age, and a value greater than 20% was taken to indicate high risk. Using the femoral neck only, the densitometric system assigned 16.4% (95% confidence interval [CI]: 15.8-17.0%) of the entire study population to the high-risk category, whereas the 10-yr absolute fracture risk system using age and femoral neck T-score classified 20.3% (95% CI: 19.7-20.9, p<0.0001) as high risk. When minimum T-score was used, the rates of high risk were similar using both approaches (31.4% [95% CI: 30.7-32.1] with the densitometric system vs 30.9% [95% CI: 30.2-31.6] for the 10-yr fracture risk system, p>0.2). A 10-yr absolute fracture risk cutoff of 20% produced the best overall agreement with the densitometric systems. Below age 65 yr, the 10-yr fracture risk system resulted in a lower rate of high-risk categorization than the densitometric system. The profile reversed after age 65 yr, with a greater proportion considered high risk in the 10-yr fracture risk system compared to the densitometric system.