Can we improve the reliability of the Constant-Murley score?

Hypothesis The Constant-Murley score (CMS) is one of the most used scales for shoulder dysfunction. The aim of this study is to determine whether the reliability of the CMS can be improved by enhancing the standardization of the items. Methods Two consecutive series of 55 patients with shoulder dysfunction were enrolled in a test-retest study and examined by 2 orthopedic surgeons with different levels of expertise. The following scores were measured: CMS, individual relative CMS, relative CMS, and standardized CMS. For each variable, the intraobserver and interobserver reliability was calculated. Results The less experienced observer had worse intraobserver reliability using the CMS (error, 4 points; 95% limit of agreement, 22) than the expert observer (error, 2.4 points; 95% limit of agreement, 16). The standardized CMS showed better intraobserver reliability, with an error of 0.4 points and 95% limits of agreement of 9 for the expert observer and 13 for the less experienced observer. The correction against the contralateral unaffected side and the reference population determined a worsening of reliability in both observers. Interobserver reliability showed an improvement similar to that of intraobserver reliability (systematic error, 4; 95% limit of agreement, 24) by use of the CMS and improved to 1 point when the standardized CMS was adopted (95% limit of agreement, 12). Conclusions This study showed that the standardization of the items significantly improved both the intraobserver reliability and interobserver reliability of the CMS. The level of expertise of the observer has less of an effect on reliability when the score is applied with a higher level of standardization.