Reliability of gray-scale and power Doppler ultrasound static image scoring for synovitis among Taiwanese rheumatologists

Abstract

Objective: To test the inter-observer reliability of gray-scale (GS) and power Doppler (PD) musculoskeletal ultrasound (MSUS) still images synovitis scoring among Taiwanese rheumatologists before and after a one-day basic MSUS training course. Methods: Thirty-seven rheumatologists completed the one-day basic MSUS training course. Before the training course, all participants scored synovitis semi-quantitatively for 14 GSUS and 10 PDUS still images. Likewise, all participants performed synovitis scoring for another sets of 14 GSUS and 10 PDUS still images after the training course. Inter-observer reliability was estimated by calculating the Krippendorff's alpha (α) coefficient with 95% confidence interval (CI). Results: Inter-observer reliability of US synovitis scoring was unacceptable, either before (α, 0.46; 95% CI, 0.37–0.55) or after (α, 0.42; 95% CI, 0.33- 0.50) the training course. The inter-observer reliability of PDUS synovitis scoring was higher than that of GSUS synovitis scoring, both before (α, 0.66; 95% CI, 0.60- 0.72) and after (α, 0.62; 95% CI, 0.55- 0.68) the training course. Before the training course, the inter-observer reliability among sonographers with more than three years' experience (α, 0.65; 95% CI, 0.59- 0.72) was higher than that among those with 1-3 years' experience (α, 0.48; 95% CI, 0.39- 0.56) and those with less than one year's experience (α, 0.49; 95% CI, 0.41- 0.57). Among all joint regions, the interobserver reliability of GSUS synovitis scoring was the lowest in the proximal interphalangeal (PIP) joints, but the inter-observer reliability of PDUS synovitis scoring was the highest in the PIP joints. This basic training course only significantly improved the inter-observer reliability in the PDUS synovitis scoring of the metacarpophalangeal joints and the metatarsophalangeal joints. Conclusion: The inter-observer reliability of GUUS or PDUS still image synovitis scoring among Taiwanese rheumatologists was poor. There is an unmet need to conduct an intensive training course to improve the inter-observer reliability of semi-quantitative synovitis scoring of US images, especially for GSUS images and for rheumatologists with less than three years of MSUS experience.