Reliability of radiographic assessment in hip and knee osteoarthritis

To evaluate the reproducibility of commonly used radiographic measures in hip and knee OA and to overcome certain limits of existing knowledge on their reliability from previous studies.

Three readers evaluated 100 hip joints (50 pelvic X-rays) and 100 antero-posterior and lateral knee films of a hospital-based sample of patients with radiographic OA at two time points 3 months apart. They retrospectively estimated the presence and severity of joint specific individual radiographic features (osteophyte formation and joint space narrowing at different sites, cysts, subchondral sclerosis, bony deformity and chondrocalcinosis) and two different overall scores. Within and between observer reproducibility was calculated by intra-class correlation coefficient.

At the hip joint excellent intra- as well as inter-observer reliability for superior joint space narrowing (JSN) and femoral head deformity could be demonstrated, while the assessment of medial JSN, osteophytes and acetabular sclerosis depends on the level of the investigator's experience. At the knee joint, femorotibial and patellofemoral osteophytes showed a high intra- and inter-observer reliability. Grading of JSN is highly reader dependent; rating of subchondral sclerosis and chondrocalcinosis does not seem to be reproducible enough. The overall scores showed an excellent reproducibility both at hip and knee joints.

A reliable radiographic severity grading of hip and knee OA is possible with the application of global scores and individual features, if joint specific items are selected and readers are trained enough.