Background Musculoskeletal ultrasound (MSU) is an imaging technique proved to be valid in several musculoskeletal conditions. In osteoarthritis (OA) it allows the identification of inflammation and structural damage. However, MSU is an operator dependent method and its widespread use has been hampered by questions related to the reliability of both, image acquisition and image interpretation.
Objectives The objective of this study was 1) to evaluate the interobserver reliability of knee OA according to the definitions used by the OMERACT reliability exercise of inflammatory and structural abnormalities in patients with knee osteoarthritis using ultrasound and 2) to compare the interobserver reliability on previous collected images (static reading, thereafter) versus after the acquisition and interpretation of images in real time (real time reading, thereafter).
Methods A reliability exercise based on the reading of US images was conducted by two experienced rheumatologists in MSU. A set of 59 images of both, normal and OA knee lesions were collected for the static reading. A set of 20 knees were scanned by each rheumatologist for the real time reading. Dichotomous and semi-quantitative scoring (0–3) was performed for the presence of damage on the condrosynovial margin, osteochondral margin and matrix of the throchlear cartilage, osteophytes at the lateral and medial femoral condyle and proximal tibia, medial and lateral meniscal extrusion and Baker's cyst. Interobserver reliability was calculated by the Cohen's kappa coefficient.
Results Interobserver reliability scores for the static reading were good for cartilage damage, meniscal extrusion and Baker's cyst, while they were excellent for the presence of osteophytes. The scores for the real time reading were poor to moderate for cartilage damage, osteophytes and Baker's cyst and good for meniscal extrusion. These results are shown in Table 1.
Conclusions This exercise shows that the interobserver reliability of MSU for the detection of knee OA lesions is widely different depending on the type of reading (static versus real time). Although MSU seems to be reliable for the detection of knee OA lesions, caution needs to be taken in the interpretation of published data regarding the type of reading exercise performed.
Ann Rheum Dis 2016;75:842–846.
Disclosure of Interest None declared