The objectives were to evaluate improvement in reliability of joint space width (JSW) measurements, according to the previous experience of the readers in hip osteoarthritis (OA).

Fifty radiographs in hip OA patients from a randomized controlled trial (ECHODIAH) were read 12 times by two readers with different levels of experience in radiographic interpretation. Intra-reader reliability and inter-reader reliability were assessed through calculation of the smallest detectable difference (SDD) and the intra-class correlation coefficient (ICC).

Intra-reader reliability: for the experienced reader, no significant change in SDD was observed with repeated JSW measurements. For the junior reader, the SDD decreased significantly from 0.90mm to 0.65mm (P=0.002). A plateau of SDD was reached after reading 200 to 300 radiographs, with a value close to the mean SDD of the experienced reader (0.67mm). ICCs were high (>0.88) for both the senior and the junior readers. Inter-reader reliability: no improvement in inter-reader reliability was noted with repeated measurements of JSW (SDD=0.81mm). However, no additional training sessions were performed. ICCs were high (>0.85).

This study indicated for the first time a learning curve in analyzing JSW in hip OA for an inexperienced reader. After analyzing 200 to 300 radiographs, the junior reader reached an intra-reader reliability similar to the experienced reader. Moreover, the information brought by the ICC was not sufficient, suggesting that the SDD should be systematically reported in studies evaluating OA radiographic progression, on top of the ICC, to help in interpreting the JSW reliability.