Inter-observer reliability of three different radiographic scores for adult haemophilia.

To evaluate the inter-observer reliability of radiological assessment systems for haemophilic arthropathy, three senior orthopaedic surgeons with expertise in haemophilia independently evaluated a total of 527 joint radiographs of adult haemophilia patients, without any knowledge of the clinical data. This study was the largest study to evaluate the reliability of radiological assessment systems. As for the results, the Arnold-Hilgartner staging system showed moderate reliability (kappa value: κ = 0.44, P = 0.000), and the De Palma grading system and the Pettersson scoring system showed fair reliability (κ = 0.40, P = 0.000) and slight reliability (κ = 0.12, P = 0.000) respectively. As for the reliability of the eight findings in the Pettersson scoring system, three findings, which were 'narrowing of joint space' (κ = 0.70 P = 0.000), 'irregular subchondral surface' (κ = 0.58, P = 0.000) and 'erosion of joint margins' (κ = 0.56, P = 0.000), showed substantial or moderate reliability. Other findings showed fair or less reliability. The traditional radiological assessment systems showed poor inter-observer reliability. Both progressive scales showed higher reliability than the additive scale, and the three findings in the Pettersson scoring system showed good reliability. These results suggested that the progressive scale, including the three reliable radiological findings, might be a more reliable radiological assessment system.