Assessment of osteopenia from spine radiographs using two different methods: the Chingford Study

Two methods for diagnosing radiological osteopenia in thoracic (TS) and lumbar (LS) spine radiographs were assessed: a subjective conventional method (A) and a semiquantitative method (B), by comparing them with bone mineral density (BMD) measured by dual energy X-ray absorptiometry (DEXA), in a population of "normal" women aged 45-70 years (n = 818). For both methods there was good intraobserver and interobserver reproducibility. BMDs were significantly lower with increasing radiological osteopenia grades (p < 0.001), and remained lower after adjustment for age and body mass index (p < 0.01). The proportion of subjects with DEXA-defined osteoporosis rose with increasing radiological osteopenia grades for both methods. The worst osteopenia categories identified 29.7-55.3% of women with DEXA-defined osteoporosis, compared with 6.1-11.7% in the "normal" categories. Both methods, however, showed a large degree of overlap of BMDs between the various radiological osteopenia grades. The sensitivity and specificity of method A in diagnosing osteoporosis were 45.3% and 78.4%, respectively, for the TS and 19.0% and 94.3%, respectively, for the LS. For method B the sensitivities and specificities were 8.8% and 96.1%, respectively (TS), and 10.2% and 95.6%, respectively (LS). Although both methods have poor sensitivities, "definite" or "high" grade osteopenia should be an indication for bone densitometry. The high specificities suggest that a "normal" (no osteopenia) X-ray is unlikely to have a significantly low BMD.