Intra- and inter-rater reliability of the detection of tears of the supraspinatus central tendon on MRI by shoulder surgeons

Background: The purpose of this study was to determine the intra- and inter-rater reliability of detecting full- and partial-thickness tears of the supraspinatus intramuscular central tendon on magnetic resonance imaging (MRI) by orthopaedic shoulder surgeons. Full-thickness tears of this tendon have previously been associated with the failure of nonsurgical management of rotator cuff tears. Methods: Shoulder MRIs from 40 patients entered into a prospective rotator cuff disease database were independently reviewed by two musculoskeletal (MSK) radiologists in order to determine if there was a partial- or full-thickness tear of the supraspinatus central tendon. The MRIs were randomly sorted and distributed to 16 fellowship-trained shoulder surgeons. The surgeons then similarly diagnosed each patient. After a 1-month interval, surgeons repeated the evaluation with the same set of randomly reordered MRIs. Surgeon intra- and inter-rater reliability was determined with the kappa statistic. Agreement and inter-rater reliability were also determined between the shoulder surgeons and MSK radiologists. Results: For full-thickness tears, the intra-rater reliability was excellent (0.86 ± 0.1, 95% confidence interval (CI): 0.81, 0.91) and the agreement was 93.4% ± 4.6, 95% CI: 91.1, 95.8. Inter-rater reliability for both rounds was also excellent (0.77 and 0.74). The agreement between the shoulder surgeons and MSK radiologists was 92.9% ± 3.9, 95% CI: 90.9, 94.9, and the kappa was 0.85 ± 0.08, 95% CI: 0.81, 0.89. Including partial-thickness tears resulted in agreement of 65-92% and kappa values of 0.59-0.72. Conclusion: The reliability for the MRI detection of full thickness tears of the supraspinatus central tendon among shoulder surgeons and between shoulder surgeons and MSK radiologists was excellent.