Intraobserver and interobserver agreement of Goutallier classification applied to magnetic resonance images

Background: Fatty infiltration of the muscle bellies is an important prognostic factor in rotator cuff tears. It was described initially in computed tomography examinations, and there is an ongoing debate about whether magnetic resonance (MR) is a reliable method for staging fatty infiltration. This study sought to determine intraobserver and interobserver agreement for Goutallier's classification of fatty infiltration of the rotator cuff as evaluated through MR imaging.

Materials and methods: Twenty MR examinations of the shoulder showing full-thickness tear of the supraspinatus tendon, with or without associated lesions, were evaluated by 3 radiologists with experience in musculoskeletal system imaging and 3 fellowship-trained shoulder surgeons. The evaluators classified the fatty infiltration of the supraspinatus muscle according to the guidelines proposed by Goutallier et al. After 8 weeks, they re-evaluated the examinations, without access to their previous reports. Weighted κ index values were determined for intraobserver and interobserver agreement analyses, and intraobserver agreement κ values are reported with 95% confidence intervals (CIs).

Results: The mean intraobserver agreement was 0.832 among the orthopedists (CI > 95%) and 0.741 among the radiologists (CI > 95%). Interobserver agreement was 0.8214 (evaluation 1) and 0.7231 (evaluation 2) among the orthopedists (P < .0001) and 0.6627 (evaluation 1) and 0.6067 (evaluation 2) among the radiologists (P < .0001). Intraobserver agreement was not associated with length of experience or frequency of routine evaluations.

Conclusions: When it is applied to MR images of rotator cuff musculature, Goutallier's fatty infiltration staging rubric yielded highly significant intraobserver and interobserver agreement.