Abstract
The aim of this multireader, multicenter study was to assess the inter-reader reliability of the score in the assessment of disease status and progression. The exercise involved 10 sets of metacarpophalangeal (MCP, 2nd to 5th) joints and 10 sets of wrist magnetic resonance images that were scored by experienced readers from 5 international centers. Synovitis was scored for each site using a global score (0-3). Bone abnormalities were assessed at 8 MCP joint sites and 15 wrist sites according to proportion of bone volume (0-10 for erosions and defects and 0-3 for edema). Intraclass correlation coefficients (ICC) and smallest detectable differences for synovitis, erosions, and edema were acceptable, although better for status scores than progression scores. The agreement for MCP joints was better than wrists. Limited variation in the images for some findings resulted in low ICC. Bone defects had the poorest agreement and have been omitted from new scoring recommendations. Despite limited training, multicenter readers demonstrated acceptable levels of agreement.
