Vertebral deformity as an index of osteoporosis-induced spinal fracture--an external validity construct based on bone density data

Abstract

In order to test the validity of vertebral morphometry for the assessment of prevalent vertebral osteoporotic fractures, we examined the association between vertebral deformities and bone mineral density (BMD). The study population consisted of 595 postmenopausal women and 581 men aged 50 to 82 years who participated in the baseline survey of the European Vertebral Osteoporosis Study (EVOS) in Germany and received BMD measurements by dual-X-ray-absorptiometry (DXA) at the femur and the lumbar spine. In both sexes only vertebral deformities defined by more stringent morphometric criteria (McCloskey; Eastell 4 SD criterion) were significantly and inversely related to BMD (odds ratios 1.42 to 3.21 for a 1 SD [standard deviation] reduction in BMD; p < 0.05). The strength of the association depended on the stringency of the morphometric algorithm applied, and on the site of BMD measurement. The strongest associations were observed with femoral neck BMD in women and with BMD at the lumbar spine in men. In contingency analyses between vertebral deformities and osteoporosis (WHO criteria; European young female BMD reference values), vertebral deformities proved to be highly specific by all methods, even slightly more so in men (87.8 to 97.5%) than in women (86.3 to 96.7%). The predictive value of a positive test with respect to vertebral osteoporosis reached a maximum value of about 50% in both sexes. It further increased up to 72%, when the definition of osteoporosis was based on low BMD values at either the spine or the femoral neck, but only in women. In conclusion, for both men and women the likelihood of vertebral deformities to be related to osteoporosis increases with the stringency of the morphometric method, with a similar probability of major deformities to represent vertebral osteoporotic fractures in men as in women. Nevertheless, even the most stringent morphometric criteria are not sufficiently valid instruments for the assessment of the prevalence of vertebral osteoporotic fractures in epidemiological studies.