Achieving consensus in ultrasonography synovitis scoring in rheumatoid arthritis

Abstract

Ultrasonography is sensitive for synovitis detection but interobserver variation in both acquisition and image interpretation is still a concern. The objective was to assess if a short collegiate consensus would improve inter-observer reliability in scoring of synovitis. Eight rheumatologists (Singapore) participated in a 1-day consensus meeting divided into: (i) still-image interpretation and consensus followed by; (ii) image acquisition and interpretation, according to definitions and synovitis scoring rules endorsed by Outcome Measures in Rheumatology (OMERACT) and TUI (Targeted Ultrasound Initiative). Interobserver reliability of semiquantitative scoring in B-mode, Power Doppler (PDUS) and European League Against Rheumatism (EULAR)-OMERACT PDUS composite score was assessed by intraclass correlation co-efficient (ICC). Agreement at the joint region level was calculated using prevalence-adjusted-biased-adjusted-kappa (PABAK). For B-mode still images, ICC was good at 0.75 (95% CI 0.66-0.82) while for PDUS images this was excellent at ICC=0.88 (95% CI 0.83-0.92) with ICC improving by 12% for B-mode and 13% for PDUS respectively. During image acquisition and interpretation, B-mode scoring showed ICC=0.75 (95% CI 0.66-0.84) while for PDUS the ICC was lower at 0.59 (95% CI 0.48-0.72). The ICC for OMERACT PDUS composite synovitis scoring was good at 0.77 (95% CI 0.68-0.85). At the joint level, agreement varied with PABAK being excellent in the small joints of the hands but poor to fair in the wrists, elbows, ankles and metatarsophalangeal joints, and no agreement at the knees (PABAK range -0.34 to 0.85). A consensus meeting was useful in improving interobserver variation in US synovitis scoring of still images, but image acquisition and interpretation especially in non-hand joints require further standardization.