An atlas improves interobserver agreement regarding application of the ISCD vertebral body exclusion criteria

Coexisting conditions such as osteoarthritis and compression fracture may spuriously elevate the dual-energy X-ray absorptiometry (DXA)-measured lumbar spine bone mass. To improve the diagnostic utility of lumbar spine DXA to diagnose osteoporosis, the International Society for Clinical Densitometry (ISCD) suggests excluding vertebrae affected by focal structural anomalies or unusual T-score discrepancies. However, we previously demonstrated only moderate agreement between physicians regarding vertebral body exclusion. We hypothesized that an atlas containing examples of vertebrae to exclude would improve interobserver agreement. Subsequently, we developed an interactive web-based atlas of lumbar spine DXA images with options to exclude vertebrae and compare one's answers to those derived by group consensus. Before and after review of the atlas, 5 ISCD-certified physicians applied the exclusion criteria to 90 DXA scans, recording the indications for vertebral exclusion on a standardized worksheet. After development and review of the atlas, interobserver agreement regarding vertebral body exclusion improved significantly (p<0.0001). We plotted the deviation of each physician's reported T-score vs the mean T-score for each of 90 scans, and demonstrated that the scatter from the mean is decreased after atlas review. Furthermore, correlations in T-score improved in 7 of 10 physician pairs after atlas review. We conclude that an interactive atlas promotes uniform lumbar spine DXA interpretation.