Vertebral dimension measurements improve prediction of vertebral fracture incidence.

We investigated the independent contributions of vertebral size and shape, bone density, and existing fractures for predicting the risk of new vertebral fractures among postmenopausal Japanese-American women in Hawaii. Baseline measurements of bone density at the distal and proximal radius and the calcaneus were obtained in 1981, and at the lumbar spine in 1984. Incident fractures were documented on serial radiographs during an average of 8.0 years of follow up of 840 women. A positive difference of 1 standard deviation in vertebral depth increased the risk of new vertebral fractures by approximately 1.3-fold (95% confidence interval = 1.03, 1.66) after controlling for bone density and prevalent fractures. A combination of large vertebral depth (above the 66th percentile) and low bone density (below the 33rd percentile) increased fracture risk approximately six times compared to women with high bone density (above the 66th percentile) and small vertebral depth (below the 33rd percentile). We conclude that combining information about vertebral depth, bone density, and prevalent fracture appears to be better for predicting new fractures than any single variable alone.