The results of different readers' interpretations of laser-digitized hand radiographs versus original radiographs were compared to determine the reproducibility of scoring of erosions (ERO), joint space narrowing (JSN), and their combination (ERO + JSN) in patients with rheumatoid arthritis (RA).
Standardized radiographs of both hands were obtained at 2 visits (baseline and 6-24-month followup) from 30 patients with established RA. Conventional and laser-digitized (pixel sizes 50 µm and 100 µm) radiographs were scored independently by 3 experienced and trained radiologists who were blinded to the order of the visits. Scoring of radiographs was based on the validated Genant grading system.
Intertechnique (intrareader) correlation coefficients at baseline were 0.90-0.93 for scoring of ERO, 0.90-0.94 for scoring of JSN, and 0.92-0.95 for ERO + JSN; for scoring of progression between baseline and followup, these values were 0.93-0.97, 0.87-0.95, and 0.93-0.97, respectively. Interreader (intratechnique) correlation coefficients at baseline were 0.82-0.96 for scoring of ERO, 0.69-0.91 for scoring of JSN, and 0.80-0.95 for ERO + JSN; for scoring of progression between baseline and followup, these values were 0.90-0.97, 0.80-0.92, and 0.90-0.95, respectively. Intrareader (intratechnique) correlation coefficients were 0.90-0.97 for scoring of the original radiographs and 0.90-0.98 for scoring of the digitized images at 100 µm.
Using this modified grading system, scoring of RA progression directly from paired, high-resolution monitors of laser-digitized images of the hands provided highly reproducible results, comparable to those obtained from the original radiographs. Thus, this method may have useful applications in clinical trials involving RA.