Determination of the minimal clinically important difference in rheumatoid arthritis joint damage of the Sharp/van der Heijde and Larsen/Scott scoring methods by clinical experts and comparison with the smallest detectable difference

Abstract

Objective. To assess the minimal clinically important difference (MCID) in joint damage on hand and foot radiographs of patients with early rheumatoid arthritis (RA) as assessed with the Sharp/van der Heijde and Larsen/Scott methods, and to study how the smallest detectable difference (SDD) relates to the MCID for each method. Methods. The judgments of an international panel of experts on the clinical relevance of progression of joint damage as seen on sets of radiographs obtained at 1-year intervals in 4 clinical settings (early versus late RA and mild versus high disease activity) were used as the external criterion, which was compared with the progression scores as determined by the 2 scoring methods. Progression scores with the highest combined sensitivity and specificity for detecting clinically relevant progression represented the MCID. Subsequently, the sensitivity and specificity of the scoring methods were determined when using the SDD as the threshold for relevant progression, and these were compared with the sensitivity and specificity of the MCID. Results. The panel judged changes in joint damage around the level of the SDD (5.0) of the Sharp/van der Heijde method as minimal clinically important, resulting in satisfactory sensitivity (mean 79%) and specificity (mean 84%) for detecting clinically important progression in the 4 clinical settings when using the SDD as the threshold value. The MCID (mean 2.3) of the Larsen/Scott method was much smaller than its SDD (5.8), and the sensitivity for detecting clinically important progression by applying the SDD as threshold was consequently low (mean 51%), accompanied by high specificity (mean 99%). Conclusion. This study suggests that the SDD of the Sharp/van der Heijde method can be used as the MCID, i.e., as the threshold level for individual response criteria. The SDD of the Larsen/Scott method, however, turned out to be too insensitive to use as the threshold for individual clinically relevant change.