A prospective, double-blind comparison of magnetic resonance imaging and arthroscopy in the evaluation of patients presenting with shoulder pain.

The purpose of this study was to determine the diagnostic ability of magnetic resonance imaging (MRI) compared with a reference standard, arthroscopy, in patients presenting with shoulder pain consistent with the signs and symptoms of shoulder impingement. Fifty-eight patients in whom conservative management failed underwent standardized MRI with intravenous gadolinium enhancement and arthroscopic evaluation. The prevalence of supraspinatus tendon, long head of the biceps tendon, and acromioclavicular joint pathology in this population was high: 79.3%, 66%, and 56%, respectively. MRI was highly accurate in detecting full-thickness supraspinatus tears and acromioclavicular joint pathology. However, it had poor concordance with arthroscopy in diagnosing pathology in the biceps tendon and in classifying the curvature of the acromion. A smaller number of other abnormal structures were identified in this population of patients, including labral abnormalities (superior labral anterior-posterior and Bankart tears), Hill-Sachs lesions, and articular surface damage. Concordance between MRI and arthroscopy would increase with improved radiologist-surgeon communication, a more reliable classification of pathology, and possibly the use of intraarticular contrast. This study demonstrates that patients presenting with the clinical findings of shoulder impingement syndrome represent a heterogeneous population. Overall, MRI is a useful tool in the identification of shoulder pathology; however, the clinical correlation of this information and the assessment of outcomes remains unknown.