[Visual evaluation of conventional roentgen pictures and dual x-ray absorptiometry in the diagnosis of osteopenia].

Routine radiographs and PA dual x-ray absorptiometry (DXA) of the lumbar spine were performed in 253 patients. A T-score of -2SD (approximately 20% reduction from young normal bone mineral density (BMD)) as measured by DXA, was used as threshold for the diagnosis of osteopenia. A lumbar spine index (LSI) as proposed by Barnett and Nordin was assessed by one radiologist. The correlation between LSI and bone mineral density, measured by DXA, was poor, and our results indicate that osteopenic patients cannot be differentiated from non-osteopenic patients by LSI. 100 radiographs of the lumbar spine were evaluated by 9 observers in order to determine observer variation in the detection of osteopenia. A complete agreement between the readers for the diagnosis of osteopenia was achieved in 43 percent of all radiographs. DXA and observers agreed in 68 to 76 percent of all cases. For a decrease of BMD of more than 60 percent compared to young normal adults, as determined by DXA, all readers agreed in the diagnosis of osteopenia, whereas for higher BMD values the agreement between readers and DXA worsened. The inconsistency between DXA and observer agreement was greatest for patients with a BMD reduction between 10 and 20 percent. The mean values of the kappa-coefficients were 0.574 +/- 0.06 for interobserver variation and 0.437 +/- 0.06 for the agreement between readers and DXA-results. An only moderate kappa-coefficient of 0.573 for the intraobserver variation as determined in one reader was found.