[Lumbar bone density. Risk factor for vertebral fractures in women]

The direct connections between a reduction in bone density and an increase in the incidence of fractures with increasing age is well known. When interpreting bone density measurements the attending physician is confronted with an overlap of false-positive and false-negative parameters, as is also the case with biochemical data. Aim of this study was to determine, in addition to bone density, a fracture threshold, i.e. a value for bone density with maximal sensitivity and specificity, in a representative cohort of women.

534 healthy women (mean age 60.9 +/- 8.07, range 40-83 years), seen in an out-patient osteological clinic between 1993 and 1996 were included. After full biochemical investigation to exclude a metabolic bone disease, bone density (LBD) of the lumbar vertebrae 1-5 and the number of vertebral body fractures were documented, together with age and menopausal status. LBD was measured by quantitative computed tomography (qCT), results being analysed with the Wilcoxon 2-sample test and logistic regression and assessed by receiver operating characteristics (ROC) analysis.

None of the women with a qCT value over 111 mg/cc had evidence of vertebral fractures. Those with values under 60 mg/cc had at least one vertebral fracture (defined as at least 15% reduction in vertebral body height). There was a highly significant negative correlation between age and LBD (r = -0.56; P < 0.0001), as well as between LBD and the number of fractured vertebrae (r = -0.58, P < 0.0001). But there was no longer any correlation between age and number of vertebral body fractures once allowance was made for LBD (r = 0.001, not significant). The least overlap of false-positive and false-negative values was at an LBD with qCT values of 102 mg/cc.

LBD is the determining variable for fracture risk independent of age. Despite an overlap of false-positive and false-negative LBD values, the probability of fractures can be predicted from bone mineral content.