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