Objectives: The aim was to assess the diagnostic performance of magnetic resonance arthrography (MRA) of the shoulder for the detection of "Fosbury flop" tears.

Methods: All patients undergoing arthroscopic rotator cuff repair were considered as potentially eligible for inclusion in this prospective case series. Preoperative MRAs were consecutively performed and compared with intraoperative findings being the gold standard control. Two experienced musculoskeletal radiologists, blinded to the arthroscopic findings, independently analysed all MRAs for the presence of typical signs of Fosbury flop tears.

Results: 423 patients were included. Among this group, 11 presented a Fosbury flop tear with a prevalence of 2.6%. Interobserver agreement was considered as excellent for thickened tendon, tendon stump, fluid accumulation, abnormal fibres orientationand adherences between tendon and bursa. The most sensitive and specific criterion was the tendon thickness[90.9% CI(62.3-98.4) and97.1% CI(95.0-98.3), respectively]. Conclusions: Fosbury flop tear is an uncommon condition that should be recognized to allow optimal anatomic repair. MRA is a valuable tool in making the diagnosing of this lesser recognized type of rotator cuff tear. An abnormally thickened supraspinatus tendon tear should especially raise suspicion for a Fosbury flop tear of the posterosuperior rotator cuff. Advances in knowledge: MRA is a valuable tool in making the diagnosing of Fosbury flop tears. An abnormally thickened supraspinatus tendon tear should especially raise suspicion for such lesion.