Abstract

To review the diagnostic performance of ultrasonography (US) and magnetic resonance imaging (MRI) for the detection of rotator-cuff tears, we performed a retrospective audit of patients who underwent shoulder arthroscopy at Hinchingbrooke hospital. The diagnostic accuracies of US for full and partial-thickness tears were 82% and 28% respectively. Those of MRI were 82% and 81% respectively. These were lower than expected from the literature. This discrepancy is likely to be the consequence of over-diagnosis in imaging and under-diagnosis at arthroscopy.