Abstract
To determine the minimal clinically important difference (MCID) between hand and foot films with a 1 year interval assessed with the Sharp/van der Heijde or Larsen/Scott scoring method. Progression scores of the 2 methods were compared with the opinion of an international expert panel on clinical relevance of radiological joint damage in 4 predefined clinical settings. The expert panel consisted of 3 rheumatologists, who evaluated 46 pairs of hand and foot films, taken with 1 year intervals, of patients with early rheumatoid arthritis. Receiver operating characteristics curves analyzed the accuracy of different threshold values (progression scores) of the 2 scoring methods to detect the presence or absence of clinically important difference, as defined by the expert panel as external criterion. The threshold value with the highest accuracy was subsequently chosen as the score representing the MCID. Five Sharp/van der Heijde units and 2 Larsen/Scott units were the best cutoffs. The accompanying sensitivities ranged from 77% to 100% for the Sharp/van der Heijde method and from 73% to 84% for the Larsen/Scott method for the 4 clinical settings. The specificities were between 78% and 84% for the Sharp/van der Heijde method and between 74% and 94% for the Larsen/Scott method. The smallest progression score that can be detected apart from interobserver measurement error, the smallest detectable difference (SDD), was equal to or larger than the calculated MCID, 5 Sharp/van der Heijde units and 6 Larsen/Scott units in our study, if the mean progression scores of the same 2 observers were used. The SDD is a conservative estimate of the MCID; our panel rated progression at or below this level as clinically significant.