Background Joint destruction in rheumatoid arthritis (RA) is comprised of hyaline cartilage and bone damage, with the former more clearly associated with irreversible physical disability than bony damage.
Objectives To test the reliability of a semiquantitative scoring system for the assessment of cartilage by musculoskeletal ultrasound (US) in a web-based exercise as well as a patient-based reliability study of patients with RA.
Methods Static images of metacarpophalangeal (MCP) joints 2–5 in RA patients and healthy controls were acquired and a dataset of 123 anonymized images including 25 duplicate images was circulated among an international EULAR-OMERACT taskforce of 25 rheumatologist experts in US who independently scored the images using a semiquantitative scoring system. Subsequently 12 taskforce members participated in a patient-based reliability study. During this meeting MCP joints 2–5 of 6 patients with RA were assessed twice on the same day by all experts using US machines (GE) equipped with high-frequency transducers (18–22MHz) with presets calibrated for the appropriate assessment of cartilage. Participants assessed metacarpal cartilage both in the standardized longitudinal midline scan as well as by freehand technique utilizing multiple planes and scored by the semiquantitative scoring system. Intraobserver reliability was assessed by Cohen's kappa and interobserver reliability by Fleiss' kappa.
Results The three-grade semiquantitative (Grade 0, normal cartilage; Grade 1, minimal change; Grade 2, severe change) scoring system demonstrated excellent (kappa: 0.87) to good (kappa: 0.73) intraobserver reliability in the web-based exercise and the patient-based reliability study respectively. Interobserver reliability was good in the web-based exercise (kappa: 0.64) and moderate (kappa: 0.49) in the patient-based reliability study. The dynamic technique performed slightly better than the longitudinal midline scan alone.
Conclusions A semiquantitative scoring system demonstrated good intra- and moderate to good inter-observer reliability in a web-based exercise and patient-based reliability study. Our study demonstrates that US is a reliable tool for evaluating cartilage and supports the use of a new semiquantitative US scoring system for evaluating cartilage change in RA.
Acknowledgements The patient-based reliability study was supported by a research grant from UCB.
Disclosure of Interest None declared