Magnetic resonance imaging for evaluation of failed repairs of the rotator cuff. Relationship to operative findings.

We used magnetic resonance imaging to evaluate thirty shoulders in twenty-nine patients who had unacceptable pain and loss of function after at least one operative procedure for a lesion of the rotator cuff. A preoperative history was recorded and physical and radiographic examinations were performed before magnetic resonance imaging. All patients had a reoperation two to 156 months (average, thirty-two months) after the operation. The results of the magnetic resonance imaging and the findings at the most recent operation were then compared with regard to the integrity of the rotator cuff, the deltoid muscle, and the long head of the biceps tendon. The sensitivity and specificity of magnetic resonance imaging for the diagnosis of full-thickness tears, partial-thickness tears, and intact rotator cuffs were 84 and 91 per cent, 83 and 83 per cent, and 80 and 100 per cent, respectively. The positive and negative predictive values were 94 and 77 per cent, 56 and 95 per cent, and 100 and 96 per cent, respectively. Three shoulders had a clinically detached and retracted origin of the deltoid muscle that was identified correctly on magnetic resonance imaging, and this finding was confirmed operatively. A rupture of the long head of the biceps tendon was identified correctly in four of six shoulders. In one of the two remaining shoulders, the rupture was obscured on magnetic resonance imaging because of a so-called balloon artefact from a nearby metal suture anchor. Magnetic resonance imaging detected two unexpected lesions: a denudation of the articular cartilage in one shoulder and a ganglion cyst in the supraspinatus muscle in another.