Shoulder ultrasonography versus arthroscopy for the detection of rotator cuff tears: analysis of errors.

PURPOSE To compare shoulder ultrasonography with arthroscopy for detecting specific rotator cuff tears. METHODS 96 shoulders in 56 men and 34 women aged 28 to 65 (mean, 42) years with rotator cuff problems underwent a shoulder ultrasonography by a single radiologist, and then a shoulder arthroscopy by a single surgeon. The ultrasonographic and arthroscopic findings were compared with regard to tendon-specific tears and disparities in diagnosis. RESULTS For detection of supraspinatus tendon tears, ultrasonography attained a sensitivity of 89%, specificity of 43%, positive predictive value of 76%, and negative predictive value of 100%. For detection of subscapularis tendon tears, ultrasonography attained a sensitivity of 30%, specificity of 100%, positive predictive value of 100%, and negative predictive value of 78%. Sensitivity, specificity, and predictive values were good in larger full-thickness tears, but were significantly reduced in sub-centimetre and partial-thickness tears, particularly of the subscapularis tendon. CONCLUSIONS Surgeons should be aware of the potential disparities between ultrasonographic and arthroscopic findings and be prepared to adjust the surgical procedure during arthroscopy.