Moderate value of non-contrast magnetic resonance imaging after non-dislocating shoulder trauma

The purpose of the present study was to determine the value of shoulder magnetic resonance imaging (MRI) obtained in the community setting interpreted by musculoskeletal radiologists in patients with shoulder pain initiated by a single non-dislocating shoulder trauma. In 56 of 61 consecutive patients who underwent shoulder arthroscopy due to pain after a single non-dislocating shoulder trauma, the data sets of non-contrast MRI were complete. These were retrospectively interpreted by three radiologists specialized on musculoskeletal MRI who were blinded for patients' history and who did not have access to the reports of arthroscopy. Standard evaluation forms were used to assess the MRIs for superior labrum anterior and posterior (SLAP) lesions, anterior or posterior labrum lesions, lesions of the long head of biceps tendon (LHB) and for partial tears of the supraspinatus tendon and the upper quarter of the subscapularis tendon. Quality of the MRI was assessed by each radiologist on a four-point scale. The pooled sensitivity for the three radiologists for the detection of SLAP lesions was 45.0 %, for anterior or posterior labrum tears 77.8 and 66.7 %, for lesions of the LHB 63.2 % and for partial tears of the supraspinatus or subscapularis tendon tears 84.8 and 33.3 %. Corresponding inter-rater reliabilities were poor (SLAP lesions) to substantial (anterior labrum tears). Quality of MRI only influenced the accuracy for the detection of posterior labrum tears. A non-contrast shoulder MRI obtained in the community setting after non-dislocating shoulder trauma has a moderate sensitivity for most intraarticular pathologies when interpreted by musculoskeletal radiologists. Accuracy is dependent on the observer and not on the assessed quality. Case series, Level IV.