Owen, Sugaya, and Hayashida Classifications Give Poor Intra- and Inter-Rater Agreement on a Magnetic Resonance Imaging Evaluation of Subscapularis Tendon Retears

To investigate the intra- and inter-rater agreements for magnetic resonance imaging (MRI) evaluations of subscapularis tendon integrity at 6 months after arthroscopic rotator cuff repairs.Patients who had isolated or combined subscapularis tears and had undergone arthroscopic rotator cuff repairs were retrospectively included. The exclusion criteria included revision of arthroscopic surgery, minor subscapularis tears without repair, and inadequate postoperative images. MRI scans 6 months after surgery were used for the purpose of accessing the integrity of the subscapularis tendons. Three orthopaedic surgeons blindly evaluated the images twice at 2-week intervals. Three currently available classifications were used: the Owen classification, the Sugaya classification, and the Hayashida classification. Dichotomization and trichotomization methods were used for the Sugaya classification and Hayashida classifications. The aforementioned classification scores were combined for the agreement evaluation. Intra- and inter-rater agreement was assessed by calculating the Fleiss' kappa coefficients.A total of 35 patients were included. Both the Owen and Hayashida classifications had poor inter-rater agreements (κ = 0.10 and 0.04, respectively) and poor-to-weak intra-rater agreements (κ = 0.27-0.44 and 0.16-0.45, respectively). The Sugaya classification had poor inter-rater agreement (κ = 0.10) and poor intra-rater agreements (κ = 0.16-0.32). Dichotomization and trichotomization of Sugaya and Hayashida classifications did not lead to superior agreements. The classification combination resulted in poor inter- and intra-rater agreements (κ = 0.01-0.12 and 0.08-0.39, respectively).The Owen classification, Sugaya classification, and Hayashida classification had poor intra- and inter-rater agreement in terms of evaluating subscapularis tendon re-tears on 6 months' postoperative MRI. The dichotomized and trichotomized classifications as well as the combined classifications from currently available classifications did not lead to superior agreements.Level IV, diagnostic: case series.