The interobserver reliability of ultrasound in knee osteoarthritis

Abstract

Objective. To assess the interobserver reliability between sonographers with different levels of experience in detecting inflammatory and structural damage abnormalities in patients with knee OA. Methods. After achieving consensus on definitions and scanning protocols, three ultrasonographers with different levels of experience in musculoskeletal US examined the knees of nine patients with OA. US examinations were conducted with independent blinded evaluations of inflammatory (joint effusion, synovial hypertrophy, power Doppler signal, Baker’s cysts) and structural (osteophytes, cortical bone irregularities, femoral hyaline cartilage abnormalities, protrusion of the medial meniscus) lesions. All abnormalities were scored by applying a dichotomous scale (0–1). In addition, at each knee joint site global scores for joint inflammation, cortical bone abnormalities and cartilage damage were calculated by summing the single-lesion scores. Reliability was assessed using kappa (κ) coefficients. Results. Seventeen knees were examined. Inflammatory abnormalities were observed with moderate to very good agreement (κ = 0.55–0.88) between the observers. From fair to very good agreement (κ = 0.31–0.82) was registered between sonographers for structural damage lesions. The overall κ was 0.716 for junior and 0.571 for beginner sonographers comparing their findings with those of senior sonographers. Conclusion. This represents the first ultrasonographic study focusing on the analysis of interobserver reliability between sonographers with different levels of experience in demonstrating inflammatory and structural abnormalities in knee OA. Globally, even considering some variable results that were mainly obtained by the evaluation of single components of bone involvement, US offered a reliable assessment of a wide set of abnormalities in knee OA.