Interpretation of lumbar spine densitometry in women with fractures.

Identification of postmenopausal women at risk of developing osteoporotic fractures is a major clinical problem. In this study the use of projected planar lumbar bone density values for individual fracture risk assessment was questioned. Osteodensitometry (DXA) results from 415 normal women, 62 women with previous vertebral compressions, and 76 women with previous low-energy fractures were analyzed, together with their body size and lumbar vertebral body size variables. The following were found: (1) Lumbar vertebral projected bone mineral areal density (BMD) and bone mineral content (BMC) of normal women correlated with body size variables (p < 0.001). (2) Lumbar vertebral body size variables also correlated with body size variables (p < 0.001). Logistic regression analysis of measured and derived physical variables from women without and with vertebral compression fractures (n = 477) showed: (3) The best compression fracture discriminator, significantly better than BMD, was BMC divided by (Hmax/165 cm)1.5 x (D/4.35 cm)1.5, where Hmax is the body height (cm) at the menopause, and D the mean lumbar vertebral diameter of the three mid-lumbar vertebral bodies (cm). This parameter was termed BMCcorr.. ROC analysis showed: (4) At a BMCcorr. true positive ratio of 80% the corresponding uncorrected BMC or BMD true positive ratio was only 60%. The corresponding false positive ratio was 6%. Lumbar osteodensitometry could not be used to identify women with a history of peripheral low-energy fractures. (5) BMCcorr. did not, unlike BMC and BMD, correlate with body size and vertebral size variables. (6) Likewise, an observed correlation between BMC and lean body mass in a subpopulation of 116 normal women was abolished when BMCcorr. replaced BMC.(ABSTRACT TRUNCATED AT 250 WORDS)