Observer variation in quantitative assessment of rheumatoid arthritis. Part I. Scoring erosions and joint space narrowing

Abstract

Six observers, including two bone and joint radiologists, two general radiologists, and two senior radiology residents, compared scores to quantitate radiographic findings in the hands and wrists of patients with rheumatoid arthritis. In the scoring system used, erosions and joint-space narrowing are graded separately. This scoring system differs from other methods in that equivocal findings are not scored, while ankylosis, subluxation, and dislocation are scored, and data from postoperative joints are included. Total radiographic scores were highly significantly correlated for all observers (R = .908-.958, P less than .001), as were subtotal scores for erosions (R = .723-.931, P less than .001) and joint-space narrowing (R = 0.843-0.966, P less than .001). Analysis of proximal interphalangeal, metacarpophalangeal, and wrist-joint scores showed highly significant correlations for all observers in each location. Highly significant correlations were found among three separate readings of two bone and joint radiologists (R = .950-.961, P less than .001). This scoring system provides highly consistent and reproducible results, even in the hands of less experienced observers.