Quantitative assessment of hand radiographs of rheumatoid arthritis: interobserver variation in a multicenter radiographic study.

Abstract

We investigated interobserver variations in the Larsen radiographic scoring method on hand radiographs of rheumatoid arthritis (RA) patients in a multicenter trial and developed a new radiographic scoring method. Thirteen experienced rheumatologists scored 10 representative RA hand radiograms with the Larsen scoring method and clarified the precipitating factors of interobserver variation. Based on this study, it was proved that the ankylotic joint, overlapping joint, and more precise erosive joint are needed for optimal radiographic evaluation. Therefore, we modified the Larsen scoring method on the basis of these precipitating factors and developed a novel radiographic scoring method. Finally, to determine which scoring system was most reliable, the interobserver variation using three methods (original Larsen method, revised Larsen method, our scoring method) were compared by 13 experienced rheumatologists and 13 residents. Our scoring method proved to have simplicity, reliability, and ease of learning. These results suggest that our novel radiological quantitative assessment method has useful applications for clinical studies in patients with RA.