Inter-rater reliability of radiographic scoringresults in rheumatoid arthritis patients

Abstract

Purpose Rheumatoid arthritis (RA) is a chronic and progressive systemic autoimmune inflammatory disease the cause of which remains unknown. The conventional radiography is widely accepted to evaluate the disease. The van der Heijde modification of Sharp technique provides separate scores for erosion and for joint space narrowing (JSN) that helps to evaluate the severity of RA. The interrater reliability in scoring same radiographs by double blind reading is estimated in this paper to help to understand the need of experienced raters in evaluating structural joint damage using the guidelines of van der Heijde modification of Sharp's grading system. Methods and Materials 29 patients (23 female, 6 male) fulfilling the classification criteria for RA according to American College of Rheumatology that have been randomly chosen took part in the study. Single films of each hand and foot were performed by standardized positioning. Totally 116 radiographs, four radiographs (two hands and two feet) of each of the 29 patients were scored blindly by two readers, both of them radiology residents in their second and third year without any previous experience in radiographic assessment of RA. Results The acquired scoring results (total 58) were assessed by analyzing the total score for erosion (29 cases) and JSN (29 cases) separately. 12 total scores (20.7%) were accounted as matching results. Acceptable differences in results were present in roughly 27.6% or 16 total scores. Unacceptably high were recorded in over 50% (51.7%) or 30 total scores. Conclusion Evaluating radiographs of RA patients using the same method by untrained raters produces great differences in the final score for each patient separately for erosion and for JSN as well. The significant difference in absolute scores most obviously results from the reader's insufficient reading experience level despite the very clearly defined scoring method.