Objectives To evaluate the interobserver reliability of ultrasonography (US) in the assessment of cartilage damage at metacarpophalangeal (MCP) joint level in patients with rheumatoid arthritis (RA).

Methods US examinations were performed on 80 MCP joints of 20 patients with RA using a MyLab70 XVG (Esaote Biomedica, Genoa, Italy), equipped with a broadband linear probe (6–18 MHz). For each patient, second and third MCP joints of both hands were examined independently on the same day by two rheumatologists (an experienced musculoskeletal sonographer and an investigator with limited US training). A multiplanar scanning technique on dorsal, lateral and volar aspects of the MCP joints was adopted. All US pathological findings were documented on at least two perpendicular scanning planes. Each joint was assessed by quadrant for the presence or absence of cartilage damage. Cartilage damage was also scored per quadrant on a five-grade semiquantitative scoring system on which investigators reached a consensus prior to the study.

Results Exact agreement between investigators was found in 173 out of 200 quadrants (86.5%) with regard to presence or absence of cartilage damage. Percentages of exact agreement for cartilage damage semiquantitative assessment at dorsal, lateral and volar quadrants were 72.5%, 52.5% and 85%, respectively, while unweighted κ values were 0.561, 0.366 and 0.766, respectively.

Conclusions The present study demonstrated moderate to good interobserver reproducibility of a semiquantitative scoring system based on qualitative morphological changes for cartilage damage at MCP joint level in patients with RA.